{ "paper_id": "J76-2004", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T02:51:36.681169Z" }, "title": "Sect ion S/ WAS NEVER PUSHED FOR PUSH S/ TRANS #70 FROM 66:VP/HEAD:O(1) TO 52:VP/VP:23(13) JUMP TRANS # 7 1 FROM 67:VP/V:0(1) TO 66:VP/HEAD:O(1) J U M P TRANS #72 FROH 68:S/AUX:0(1) TO 67:VP/V:0(1) JUMP TRANS #73 FROM 69:S/NO-SUBJ:O(1) TO 67:VP/V:0(1) JUMP TRANS #74 FROM 68:S/AUX:0(1) TO 69:S/NO=SUBJ:O(l) JUMP TRANS 875 FROM 69:S/NO=SUBJ:O(l) TO 67:VP/V:0(1) JUMP TRANS #76 FROM 70:S/THERE:O(l) TO 6?:VP/V:0(1)", "authors": [ { "first": "Bolt", "middle": [], "last": "Beranek", "suffix": "", "affiliation": { "laboratory": "", "institution": "Boston University", "location": {} }, "email": "" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "This research was principally supported by the Advanced Reeearch Projeclts Agency of the Department of Defense (ARPA Order No. 2904) and was monitored by ONR under Contract No. N0001.4-75-c-0533. Partial support of the author by NSF grant GS-39834 to Harvard University i s gratefully acknowledged.", "pdf_parse": { "paper_id": "J76-2004", "_pdf_hash": "", "abstract": [ { "text": "This research was principally supported by the Advanced Reeearch Projeclts Agency of the Department of Defense (ARPA Order No. 2904) and was monitored by ONR under Contract No. N0001.4-75-c-0533. Partial support of the author by NSF grant GS-39834 to Harvard University i s gratefully acknowledged.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "3. As a c o n s e q u e n c e of 1 a n d 2 , when an u t t e r a n c e i s scanned t o t r y t o i d e n t i f y t h e w o r d s , i t i s r e a s o n a b l e t o s u p p o s e t h a t a n u m b e r of ( p e r h a p s o v e r l a p p i n g ) c a n d i d a t e s w i l l be f o u n d .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "T h l s is i l l u s t r a t e i n F i p u r e 1 . 1 by a s t r u c t u r e c a l l e d a word l a t t i c e w h i c h shows schematically t h a t many Words may i n i t i a l l y appear t o be present. I n t h i s r e p r e s e n t a t i o n , t h e numbers along t h e h o r i z o n t a l s c a l e a r e se~ment boundary p o i n t s i n t h e u t t e r a n c e which r o u p h l y correspond t o points i n time A Word L a t t i c e Sentence: Give me a l l glass samples ' w i t h magnetite.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "lead not", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "In the system described here, such a word l a t t i c e can be been . did represented by a c o l l e c t i o n of word matches, each of which is composed of a word, t h e boundary points a t t h e left a n d r i g h t , ond I ends of t h e portion of the u t t e r a n c e where i t appears t o match y@ well, and a score i n d i c a t i n g how well i t matches t h e i d e a l ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "c o m b i n a t o r i a l e x p l o s i o n of s y n t a c t i c a l l y c o r r e c t b u t m e a n i n g l e s s s u b s e t s of t h e u t t e r a n c e . Even i n t h e small word l a t t i c e o f F i g u r e 1 . 1 It c a n b e s e e n t h a t t h e r e a r e n u m e r o u s s h o r t s e q u e n c e s w h i c h a r e s y n t a c t i c a l l y b u t n o t s e m a n t i c a l l y v a l i d (tr .g.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S y n t a x m u s t i n t e r a c t w i t h s e m a n t i c s i n o r d e r t o c u t down t h e", "sec_num": "3." }, { "text": "\"Ten p e o p l e a r e g l a s s s a m p l e s w i t h m a g n e t i t e f f , \"glas-s samples g i v e m a g n e t i t e w , t l l u n a r samples g i v e m a g n e t i t e t f , \" s a m p l e s g i v e l e a d n ,", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S y n t a x m u s t i n t e r a c t w i t h s e m a n t i c s i n o r d e r t o c u t down t h e", "sec_num": "3." }, { "text": "\" p e o p l e a r e p e r c e n t i f , e t c . ) . i n c o r r e c t , b u t t h a t g a p s may a p p e a r i n t h e i n p u t a s w e l l . [ I , 2 , 7 , 8 , 1 6 , 1 9 , 2 0 , 2 1 , 2 2 , 28, 2 9 , 33, 351.", "cite_spans": [ { "start": 144, "end": 159, "text": "[ I , 2 , 7 , 8", "ref_id": null } ], "ref_spans": [], "eq_spans": [], "section": "S y n t a x m u s t i n t e r a c t w i t h s e m a n t i c s i n o r d e r t o c u t down t h e", "sec_num": "3." }, { "text": "S i n c e SPARSER was i m p l e m e n t e d a s p a r t o f a s p e e c h u n d e r s t a n d i n g s y s t e m c a l l e d SPEECHLIS w h i c h i s u n d e r d e v e l o p m e n t a t B o l t b e r a n e k a n d Newman T h e o v e r a l l d e s i g n of t h e s y s t e f n i s i l l u s t r a t e d i n F i g u r e", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The i n p u t t o a", "sec_num": "4." }, { "text": "2 . 4 .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The i n p u t t o a", "sec_num": "4." }, { "text": "T h e a c o u s t i c s c o m p o n e n t a n a J y z e s t h e a c o u s t i c s i g n a l t o e x t r a c t f e a t u r e s a n d s e g m e n t t h e u t t e r a n c e i n t o a i a t t i c e o f a l t e r n a t i v e p o s s i b l e s e q u e n c e s of p h o n e m e s t~c h w a r t z a n d Makhoul ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "The i n p u t t o a", "sec_num": "4." }, { "text": "T h e s e m a n t i c c o m p o n e n t u s e s a s e m a n t i c n e t w o r k t o a s s o c i a t e s e m a n t i c a l l y r e l a t e d w o r d s a n d t o j u d g e t h e m e a n i n g f u l n e s s of a h y p o t h e s i z e d i n t e r p r e t a t i o n (See Nash-Wabber S e c t i o n 3", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SPARSER.", "sec_num": null }, { "text": "The Grammar W e have chosen t h e Augmented T r a n s i t i o n Network formalism", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SPARSER.", "sec_num": null }, { "text": "[32] f o r t h e grammar which d r i v e s SPARSER because i t is a r e p r e s e n t a t i o n which allows merging of common p o r t i o n s of the a n a l y s i s , i t is amenable t o both bottom u p and top down parsina techniques, i t f a i r l y c l e a r l y s e p a r a t e s t h e use of l o c a l information from infoflmation which was obtained from a d i s t a n t portion of t h e u t t e r a n c e and, t h e author s previous experience input s t r i n g -information which would normally be inaccessible P a g e 11 b e o a u s e i t would be h i d d e n on t h e s t a c k d u r i n g t h e p a r s i n g of s u b -c o n s t i t u e n t s .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SPARSER.", "sec_num": null }, { "text": "T h e r e a r e a p v e r a l r e a s o n s f o r n o t a l l o w i n g t h i s m e c h a n i s m i n t h e speech p a r s e r S u p p o s e , i n t h e i n p u t t h a t 1-oaks l i k e I f . . . i t would become a n i n t e g r a l p a r t of t h e a n a l y s i s a t t h e lower l e v e l , a n d i f a n o t h e r word were t o b e h y p o t h e s i z e d i n i t s p l a c e , t h e lower l e v e l t h e a n a l y s i s would h a v e t o be r e d o n e even i f none of t h e words i n t h e r e l a t i v e c l a u s e had b e e n ch'anged. Two grammars w h i c h w i l l f i g u r e p r e d o m i n a n t l y i n the r e m a i n d e r of t h i s p a p e r have b e e n w r i t t e n i n t h e MATN formalism.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SPARSER.", "sec_num": null }, { "text": "PAST VP V G I V", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "One i s a n e x t e n s i v e grammar w h i c h c a n h a n d l e ,many q u e s t i o n s , Exampled are g i v e n below which ware produced using this pramnar. I n o r f o r t h e parser d e r t o move from r i g h t t o l e f t ( t o p r e d i c t w h a t c o u l d p r e c e d e t h a t f i r s t given w o r d ) , i t must be able to d e t e r m i n e for a n y s t a t e which a r c s can enter i t , a n d f o r any a r c which s t a t e i t comes f r o m . for a n y s t a t e o n e can determine w h a t a r c s l e a v e i t a n d f o r a n y arc (except POP) o n e c a n d e t e r m i n e which s t a t e i t t e r m i n a t e s o n , 3 ) They need not ( a n d g e n e r a l l B e g i n n i n g i n s t a t e NP/ of t h e grammar p a r s e r b e i n g i n a g i v e n s t a t e ( s a y NP/QUANT) a t a given p o i n t i n t h e u t t e r a n c e ( s a y 1 4 ) . W e w i l l w r i t e c o n f i g u r a t i o n s as STATE:POSITION i n t e x t ( e . g . NP/QUANT:14) a n d s c h e m a t i c a l l y as a box w i t h i n y h i c h a r e w r i t t e n t h e s t a t e a n d t h e p o s i t i o n . appropriate l i s t , and when we t a l k of adding a n ax is tin^ configuration o r t r a n s i t i o n t o a theory we mean a d d i n g i t t o the npprbpriate l i s t . Therefore, removing a c o n f i~u r a t i o n o r t r a n s i t i o n from a theory means removing i t from the s y n t a c t i c part of the thebry, not removing i t e n t i r e l y from SPARSER s data base.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "Like c o n f i~u r a t i o n s , t r a n s i t i o n s are unique, s o only one t r a n s i t i o n i s ever constructed from point A t o point B f o r a r c X and i n p u t Y . W e w i l l frequently speak of c r e a t i n g a t r a n s i t i o n o r a configuration, b u t the reader must bear i n mind t h a t i f such a c o n f i~u r a t i o n or t r a n s i t i o n already e x i s t s , t h i s f a c t w i l l be recognized and t h e pre-existing configuration or t r a n s i t i o n w i l l be used. (Timing, measurements i n d i c a t e t h a t it takes about ,052 seconds t o c r e a t e a configuration and only .01 seconds t o t e s t i f a p a r t i c u l a r configuration already e x i s t s .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "For t r a n s i t i o n s , creation takes about .54 seconds and recognition .012 seconds.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "The sequence of configurations and t r a n s i t i o n s which would parse t h e above example is displayed i n Figure 4 .3.", "cite_spans": [], "ref_spans": [ { "start": 112, "end": 120, "text": "Figure 4", "ref_id": "FIGREF22" } ], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "A conpected sequence of t r a n s i t i o n s and configurations is c a l l e d a a t . If the sequence begins w i t h an i n i t i a l configuration and ends w i t h a t r a n s i t i o n representing a POP a r c , POP NIL bean b e pre-calculated by t h e grammar indexing package so the appropriate configurations and t r a n s i t i o n s may be constructed.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "These t r a n s i t i o n s ape cplled lead-in t r a n s i t i o n s . T h u s t h e map becomes t h a t i n of l e f t context could be a p p r o p r i a t e . For example, th'e CAT ADJ a r c in MINIGRAMMAR which e n t e r s s t a t e NP/QUANT implies t h a t an ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "S REL S NP PRO I FEATS NU SG AUX TNS", "sec_num": null }, { "text": "free t e s t on t h e a r c , a n d t h e C u r r e n t i n p u t ) t o d e t e r m i n e w h e t h e r a t r a n s i t i o n c a n be made to e x t e n d If the new word ( o r wokds) i n a theory a r e a t the and (or i n the middle) of an i $ l a n d , when SPARSER begins t o parse t h e i s l a n d i t w i l l discover t h e e x i s t i n g configurations and t r a n s i t i o n s from t h e previous theory. Whenever a t r a n s i t i o n which can be used i n t h e current theory i s discovered i n t h e map, i t and i t s 'terminating configuration are added t o the s y n t a c t i c p a r t of the current theory. T h i s i s c a l l a d t r a c i n q the t r a n s i t i o n . c r e a t i o n of a t r a n s i t i o n i s i n d i c a t e d by n a m i n g t h e t y p e of a r c causing t h e t r a n s i t i o n , the ( u n i q u e ) number of t h e t r a n s i t i o n , a n d t h e c o n f i g u r a t i o n s a t each a n d of t h e t r a n s i t i o n . ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "T h e grammar a s s o c i a t e s w i t h t h e s t a t e of t h e c o n f i g u r a t i o n a l i s t of a r c s w h i c h m a y be t e s t e d ( u s i n g t h e a r c t y p e , t h e c o n t e x t", "sec_num": null }, { "text": "[ O n l y t h i s o n e is chosen because i t h a s a h i~h a r s c o r e t h a n c o n f i g Q r a t i o n 2 , s i n c e t h e u s e of a noun a s a h e a d n o u n of a n o u n phraac! is more l i k e l y t h a n i t s use au a TRYING POP ARC POP TRANS # 1 1 FROM 11:PP/NP:21 7SELECTED CONPIGS (8) FOR EXTENSION PICKING UP .CONFIG 8:NP/ADJ:21(5) STARTING AT 21:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "MONITORING [ N ] MONITORING [ N 1 ALL ARCS TRIED AT THIS C O N F I G", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "(Now the t h e w y has been procdssed. ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "MONITORING [ N ] STARTING AT 2 1 , SCORE 1 0 MONITORING r PREP ] ENDING AT 12, SCORE 10 MONITORING [ N ] ENDING AT 1 2 , SCORE 1 0 MONITORING [ QUANT ] E N D I N G AT 1 2 , SCORE 1 0 MONITORING [ A D J ] ENDING AT 1 2 , SCORE 10 MONITORING [ ART 1 ENDING AT 12,' SCORE 1 0 PROPOSING (PREP QUANT A R T ) ENDING AT 1 2 FINISHED THEORY 2 WITH SYN SCORE 10", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "{ T h e processing of t h i s t h e o r y tbok approximately 4 . 5 seconds. )", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "T h i s example has shown t h e t r a c e produced by runninp SPARSER T h i s example uses t h e", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "SPEECHGRAMMAR described i n [4].", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "The u t t e r a n c e \"What is t h e r e g i s t r a t i o n fee?\" was s p o k e n by an a d u l t male s p e a k e r i n a quite room a n d was record on t a p e .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SELECTED CONFIGS (6) FOR EXTENSION", "sec_num": null }, { "text": "warnentat i o n a n d l a b e l i n r r o u t i n e s of t h e BBN speech u n d e r s t a n d i n s ~y s t e m . The i n i t i a l s c a n o f t h e u t t e r a n c e , u s i n p", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Tha u t t e r a n c a was a u t o m a t i c a l l y d i q i t i z e d rind p a s s e d t h r o u g h t h e", "sec_num": null }, { "text": "The two best matches, for \"what1! and ftrzgistrationll, appear to be good candidates for a theory, so we begin by build in^ and procasring that theory. NOTICING ( 4 FEE 1 9 23 1 5 5 Q), SCORE -5 NOTICING ( 9 THE 4 6 1 0 3 0 1 , SCORE 0 PROPOSING (ONLY WHOSE) ENDING AT 6 PROPOSING (ZERO NO POINT A WHOSE WHO WHICH THAT WHOM THERE) STARTING AT 3 (27 26 12 1 1 10 9 22 2 5 There a r a many places in the SPEECHGRAFIMAR which push for noun phrases, and since there were no monitors in the WFST which can us& thiv constitusnt, all of them must be tried, resulting in a PUSH NP/ TRANS 1 3 1 FROM 40:R/WH:4(1) TO 39:S/NP:23(9) ENDING AT 4:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Tha u t t e r a n c a was a u t o m a t i c a l l y d i q i t i z e d rind p a s s e d t h r o u g h t h e", "sec_num": null }, { "text": "{MONITORING [ PREP ] STARTING AT 1 9 , S C O R E 0 MONITORING [ WHOSE WHO WHICH THAT WHOM THERE ] STARTING AT 1 9 , SCORE 5 MONITORING [ A D J N V ORD QDET A R T QUANT POSS ] ENDING AT 6 , SCORE 0 MONITORINC [ WHOSE ONLY ] ENDING AT 6 , SCORE 5 MONITORINC [ MODAL V INTEGER NPR N A D J V ADV PREP ] STARTING AT 3 , SCORE 0 MONITORING [ WHOSE WHO WHICH THAT WHO11 THERE ZERO NO POINT A ] STARTING AT 3 , SCORE 5 ) PROPOSING ( V", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "ARSER PROCESSING", "sec_num": null }, { "text": ") B E G I N N I N G A T T R A N S 2 2 , C O N F I G 15 +*++ MADE # l F R O M 4 T O 23: NP DET A R T THE A D J NP N R E C I S T R A T I O N NU SC N FEE F E A T S", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "ARSER PROCESSING", "sec_num": null }, { "text": "MONITORING [ R/WHOSE 1 MONITORING [ WHICH THAT WHO WHOM WHICH WHOM ]", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "ARSER PROCESSING", "sec_num": null }, { "text": "{PROPOSING \"THATf1 llWHO1l llWHOMu \"WHICH\" \"WHOM\") {There arc two arcs e n t w i n e s t a t e R/WH w h i c h use t h e words l'whichn and llwhomv. There is a c h e c k made t o see that d u p l i c a t e p r o p o s a l 8 alre not a c t u a l l y communicated t o the control c o m p o n e n t , a l t h o u g h they appear t o be d u p l i c a t e d in the trace. }", "cite_spans": [], "ref_spans": [ { "start": 182, "end": 353, "text": "that d u p l i c a t e p r o p o s a l 8 alre not a c t u a l l y communicated t o the control c o m p o n e n t , a l t h o u g h they appear t o be d u p l i c a t e d", "ref_id": null } ], "eq_spans": [], "section": "ARSER PROCESSING", "sec_num": null }, { "text": "PUSH NP/ TRANS #32 FROM 41:S/DCL:4(1) TO 39:S/NP:23(9) JUMP TRANS #33 FROM 42:S/:4(1) TO 41:S/DCL: ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "NP/ WAS NEVER PUSHED FOR", "sec_num": null }, { "text": "{MONITORING [ NP/ NP/ V V ADV V ] }", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "NP/ WAS NEVER PUSHED FOR", "sec_num": null }, { "text": "{NOTICING \"IS\" \"ARE\" \"IS\" \"AREr1} {The words \"isV1 and f a i l e d the? context f r e s test on the arc causing tha last monitor, h s n c e they are not noticed. } JUMP TRANS 839 FROM 45:S/AUX:4(1) TO 49:VP/V:4(1) JUMP TRANS /I40 FROM 43:S/NO-SUBJ:4(1) TO 49:VP/V:4(1) JUMP TRANS #41 FROM 43:S/NO=SUBJ:4(1) TQ 49:VP/V:4(1) JUMP TRANS 842 FROM 50:S/THERE:4(1) TO 49:VP/V: ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "NP/ WAS NEVER PUSHED FOR", "sec_num": null }, { "text": "The feature NPU can be tested by tne semantic component to determine that the constituent is a noun phrasa utterance. I f nscessary, it could also be tasted on a PUSH S/ arc in the prammar; since there are aome times , e .~. d u r i n~ the construction of a s e n t a n t i a l complement, when an embedded sentence must contain a verb.) SYN WEIGHT + SEM WT = 10 + 0 = 10 {No arcs in this ramm mar push for noun phrase utterances, so this constituent is not used f u r t h a r . ", "cite_spans": [], "ref_spans": [ { "start": 451, "end": 478, "text": "is not used f u r t h a r .", "ref_id": null } ], "eq_spans": [], "section": "is an example of a constituent w h i c h h a s features attached t o i t .", "sec_num": null }, { "text": "V V V ADV PREP V 1 ENDING AT 4, SCORE 10 MONITORING [ MODAL V INTEGER NPR N ADJ V ADV PREP 1 STARTING AT 3, SCORE 0 MONITORING [ ZERO NO POINT A WHOSE WKO WHICH THAT WHOM THERE] STARTING A t 3, SCORE 5) PROPOSIHG (MODAL) FROM 3 TO 4 PROPOSING (MODAL PREP) STARTING AT 3 PROPOSING (PREP MODAL NEG QADV) E N D I N G AT 4 CREATING THEORY 3: 0 WHAT 3 4 THE 6 REGISTRATION 19 FEE 23 WITH SYN SCORE 15", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "is an example of a constituent w h i c h h a s features attached t o i t .", "sec_num": null }, { "text": "{Tkis $ v e n t took 34.5 saconds, l a r g e l y because of the ex enslvz b o t t o m up procdssinq n e c e s s i t a t e d by t h e discovery of t h e noun phrases w h i c h were not monitored for. 1 (Processing t h e proposals from t h i s t h e o r y results in the b a v t event being t h e one f o r \" i s w i n the last g a p . The word \"arevt a l s o fills the gap, b u t t h e lower l e x i c a l acore p r e v e n t s t h s event for i t from s u r f a c i n~. If i t were syntactically procasued, however, n o new t h e o r y would be created since t h a completed s t r i n g would ", "cite_spans": [], "ref_spans": [ { "start": 201, "end": 597, "text": "(Processing t h e proposals from t h i s t h e o r y results in the b a v t event being t h e one f o r \" i s w i n the last g a p . The word \"arevt a l s o fills the gap, b u t t h e lower l e x i c a l acore p r e v e n t s t h s event for i t from s u r f a c i n~. If i t were syntactically procasued, however, n o new t h e o r y would be created since t h a completed s t r i n g would", "ref_id": null } ], "eq_spans": [], "section": "is an example of a constituent w h i c h h a s features attached t o i t .", "sec_num": null }, { "text": "{This took 30.8 seconds.) {Now we w i l l process t h e c o n s t i t u e n t e v e n t for t h e theory just created. e a c h c o n t e x t f r e e p a t h t h r e u~h a n i s l a n d a n d w a l k i t i n a s p e c i a l mode a f t e r t h e i s l a n . .. two-fifty for . . .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "C R E A T I N G THEORY 3: 0 WHAT 3 4 T H E 6 R E G I S T R A T I O N 19 23 WITH SYN S C O R E 15", "sec_num": null }, { "text": ". . . two-fifty-four . , . ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "b", "sec_num": "5" }, { "text": "t h e MATN formalism, t h e t e s t component of each a r c i s , on a l l b u t t h e PUSH a r c , a l i s t of two t e a t s . The f i r s t i s a t e s t on t h e", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "s t a t e w h i c h c a n b e g i n t h e p a r s i n g of a c o n s t i t u e n t ) , i t i s c a l l e d", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "t h e l e x i c a l r e t r i e v a l c o m p o n e n t , p r o d u c e d a w~r d l a t t i c e of fifteen entries, i n c l u d i n~ s e v e r a l f o r i n f l e c t i o n a l e n d i n g s . ( I n t h i s v e r s i o n of t h l system, t h e y were not combined with the root f o r m i n t o a s i n g l e word h a t c h , a n d hsncr? c o u l d m a t c h e v a n w i t h o u t a r o o t w o r d . ) Tha format f o r a w o r d match is: (NUMBER WORD LEFT-END RIGHT-END LEXICAL-SCORE). (2 WHAT 0 3 191) ( 3 ONE 0 3 189) ( 1 1 WHEN 0 3 102) (9 THE 4 6) ( 1 REGISTRATION 6 19 237) ( 1 0 REGISTRATION 7 19 1 0 3 ) (5 HAS 9 12 121)", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [ { "text": "JUMP TRANS #80 FROM 74:S/NO-SUBJ:3(1) TO 72:VP/V:3(5$ J U M P TRANS #81 FROM 73:S/AU~:3(1) TO 74:S/NO-SUBJ:3(1) JUMP TRANS 882 FROM ~~sS/NO-SUBJ:~(I) TO 72:VP/V:3(5) JUMP TRANS 883 FROM 75:S/THERE:3(1) TO 72:VP./V:3(5) Vz now ahow how t h e same u t t e r a n c e u s e @ i n t h e p r d v i o u s example cah be recognized when d k f f e r z n t t h e o r i e s qre c r s a t e d a n d when a v a n t s a n d t h e o r i e s are processed i n a d i f f a r e n t orda? from t h a t i n Example 2 . Suppose t h a t a f t e r t h e i n i t i a l s c a n o f the P t t e r a n c e the s e m a n t i c c o m p o n e n t created two t h z o r i a s , one f o r t h e w o r d s llwhatw a n d \" f e e f f a n CREATING", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "annex", "sec_num": null }, { "text": "theory . It tttracesll them all, that is, it does not recreate them but s i m p l y p u t s the transition numbers on a l i s t which w i l l form part o f the syntactic infornation associated w i t h the current theory. The tracing process a l s o involves the creation of monitors (and notices, uherr applicable) for constituants along the path. These monit~rs and notices must be remade, since the previous monitors will activate only t h 6 previous t heary .Due to the recursive nature of the tracin~ process, t h e transitions a not necessarily followad in t.he same order t h a t t h e y were originally c r e a t e d , nor are the monitors made in exactly t h e same order.Notice that the many arcs which w e tried but which did not result in the creation of transitions in t h e previous theory are not retried h e r e . } PICEING UP CONFIG 1 : S / : 0 ( 1 ) WITH WORD WHAT TRAGING JUMP S/Q TRANS 1 FROM I : s / : o ( I ) TO 2 :~/~: 0 ( 6 ) NONITORING [ NP/ 1 TRACING CAT QWORD TRANS 2 USING \"WHATu FROM 2:S/Q:0(6) TO 4:S/NP:3(11) STARTING AT 3:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "STARTING AN ISLAND STARTING AT LEFT END OF SENTENCE SELECTED CONFIGS ( 1 ) FOR EXTENSION {Upon picking u p this configuration to extend it, SPARSER finds t h d transitions which were created d u r i n~ the processing of the word I1whatfr by tha previous", "sec_num": null }, { "text": "TO 5:MP/ORD:3(16) STARTING AT 3:", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ": S / N P : 3 ( 1 1 ) TRACING CAT QDET T R A N S 3 U S I N G \" W H Q 1 I FROM 3 : N P / : 0 ( 1 1 )", "sec_num": null }, { "text": "[This does not mean that confiauration 6 was just created.Since it already existed in the map, having been craated during the p~ocessinc of the previous theory, the configuration number is merely put on the list of configurations in the current theory.)TRACING JUMP N P / Q U A N T TRANG 5 FROM 5 :~p / O R D : 3 ( 1 6 ) TO 7:NP/QUANT:3(21) TRACING JUMP NP/DET TRANS 6 FROM 7 : N P / Q U A N T : 3 ( 2 1 ) TO 8:NPIDET:3(26) STARTING AT 3 : {NOTICING 'WHOSE\" \"WHOw ) SETTING UP CONFIG 1 2 : P P / : 3 ( 2 9 ) SETTING UP CONFIG 13:R/NIL:3(29) {No proposals were made here because proposals are not theory dependent; that is, t h e word proposals w h i c h were made during t h e processing of t h e previous theory resulted in some words baing placed in the word lattice whLch were noticed here. Remaking the proposals would n o t lead t o t h e discovery of any new information.) TRACfNG POP TRANS 8 FROM 1 0 : N P / H E A D : 3 ( 2 9 ) {The processinp of t h z i s l a n d for l l r e g i s t r a t i o n w is i d B n t i c a l to t h a t i n t h e l a s t e x a m p l e , so t h e remainder of t h a t r a c e w i l l be o m i t t e d . ENDING AT 4:NP/ WAS NEVER PUSHED FOR PUSH NP/ TRANS 136 FROM 47:VP/HEAD:4(1) TO 48:VP/NP:19(7) ENDING AT 4:NOTICING \"PAYn JUMP TRANS 837 FROM 49:VP/V:4(1) TO 47:VP/HEAD:4(1) ENDING AT 4:(NOTICING \"IS1' \"ARE\" \"PAY\" l'IS\" \"AREt1 \"PAY\") JUMP TRANS 138 FROM 45:S/AUX:4 1 ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "SETTING UP CONFIG 6 : Q U A N T / : 3 ( 1 6 )", "sec_num": null }, { "text": "I } T h e syntactic categories: ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "{MONITORING NP/ N -Q D E T ADJ INTEGER A R E Q U A N T PRO NPR POSS V V ADV TEST(N0T (CAT V))", "sec_num": null }, { "text": "T h s syntactic features: ( 1 9 7 3 ..", "cite_spans": [], "ref_spans": [ { "start": 26, "end": 35, "text": "( 1 9 7 3", "ref_id": null } ], "eq_spans": [], "section": "(.V 85 (AFFORD AM ANTICIPATE A R E A R R A N G E ASK ASSUME ATTEND BE BEEN BEGINNING B E I N G BUDGET CAN CANCEL CHANGE CONTINUE COST C O S T I N G COSTS DID DO DOES D O N E E N D ESTIMATE-V EXPECT FIGURE GET GETS GETTING G I V E GIVEN GIVES G I V I N G GO GOES GOING GONE GOT G O T T E N HAD HAS HAVE H A V I N G I S KNOW LAST LEFT LIST MADE MAKE MAKES MAKING NEED NOTE PAY PRINT PROJECT-V REMAIN REVISE SCHEDULE SEND S E N D I N G SENDS SENT SPEND SPENDING SPENDS SPENT START SUPPOSE TAKE TAKEN TAKES TAKING TOOK TOTAL TRAVEL V I S I T WANT W A S WENT WERE W I L L ) )", "sec_num": null }, { "text": "R o v n z r , P . , Nash-Wabber, B . , a n d Woods, W. ( 1 9 7 4 ) C o n t r o l C o n c e p t s i n a S p e e c h U n d e r v t a n d i n g System I E E E T r a n s . on ASSP,, 3 2 : 1 ( F a b : 1 9 7 5 ) , p . 1 3 6 -1 4 0 Wolf, Jared ( 1 9 7 6 ) S p a a c h R e c o g n i t i o n a n d U n d e r s t a n d i n g i n Fu, S y n t a c t i c P a t t e r n R e c o g n i t i o n[32] Woods, W.A-1 (-193Q) T r a n s i t i o n N e t w o r k Grammars for N a t u r a l L a n g u a g e A n a l y s i s CACM, 13:10 (Oct 1 9 7 0 ) p. 591'9606 1331 Woods, W e A w ( 1 9 7 4 ) M o t i v a t i o d a n d O v e r v i e w of BBN'SPEECHLIS.: An E x p e r i m e n t a l P r o t o t y p e far S p e a c h U n d a r s t a n d i h g Rassarch I E E E Trans. o n ASSP, 23:1 ( F z b . 1 9 7 5 ) , p . 2-10[34] Woods, W . A . ( 1 9 7 5 ) S y n t a x , S e m a n t i c s , a n d Speach ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "DOES DOES GETS G I V E S GOES H A S I S MAKES SENDS SPENDS TAKES", "sec_num": null } ], "bib_entries": {}, "ref_entries": { "FIGREF0": { "text": "e r s t a n d i n g s p e e c h i s a n e x t r e m e l y c o m p l e x process w h i c h r e q u i r e s t h o use of many t y p e s of k n~w l e d e a , o n e of w h i c h i s s y n t a x . This r e p o r t p r e s e n t s a s y s t e m c a l l e d SPARSER w h i c h is d e s i g n e d t o p r o v i d e a n d use t h e s y n t a c t i c k n o w l e d g e n e c e s s a r y t o s u p p o r t an a r t i f i c i a l s p e e c h u n d e r s t a n d i n g system. (We w i l l assume f o r t h e r e m a i n d e r of t h i s p a p e r t h a t u n l e s s e x p l i c i t l y s t a t e d o t h e r w i s e \" s p e e c h w m e a n s grammatical s p e e c h s p o k e n a t a mo4erate r a t e w i t h n a t u r a l i n f l e c t i o n s a n d p a u s e s , s p o n t a n e o u s l y p r o d u c e d b u t s i m i l a r t o t h e t y p e of s p e e c h p r o d u c e d b y r e a d i n g r a c t e r i s t i c s of s p e e c h a n d a s p e e c h p r o c e s s o r : 1 . T h e r e is n o t e n o u g h i n f o p m a t i o n i n t h e s p e c h s i g n a l t o u n i q u e l y i d e n t i f y t h e p h o n e m e s o r w o r d s i n a n o r m a l l y s p o k e n u t t e r a n c e . 2. The a c o u s t i c p r o c e s s i n g c o m p o n e n t of any a r t i f i c i a l s p e e c h u n d e r s t a n d i n g system w i l l i n t r o d u c e a d d i t i o n a l e r r o r s a n d a m b i g u i t y as i t a t t e m p t s t o i d e n t i f y t h e p h o n e m e s -o r w o r d s i n t h e u t t e r a n c e .", "num": null, "type_str": "figure", "uris": null }, "FIGREF1": { "text": "t t i c e was prodused by t h e l e x i c a l r e t r i e v a l component of t h e BBN speech understanding system from an u t t e r a n c e which h a d b e e n begmented a n d l a b e l e d by hand under conditions d e s i p n e d to simulate the performance of an automatic sepmrntar a n d l a b c l e r .", "num": null, "type_str": "figure", "uris": null }, "FIGREF2": { "text": "p r e s e n t a t i o n o f t h e word, We a l s o make a number of a s s u m p t i d n s a b o u t t h e n a t u r e of t h e s p e e c h u n d e , r s t a n d i n g p r o c e s s a n d t h e charactelristic,s o f a s y s t e m t o c a r r y o u t t h a t p r o c e s s : 1 . P e o p l e c a n u n d e r s t a n d ua s p e a k e r e v e n when t h e s p e e c h . i s f a i r l y u n g~a m m~t i c a l , s o a s y n t a x -d r i v e n s y s t e m w h i c h w o u l d a c c e p t o n l y i n p u t marbing r i g i d s y n t a c t i c r e q u i r e m e n t s w o u l d n o t be a d e q u a t e f o r n a t u r a l , c o n v e r s t i o n a l s p e e c h . 2. S i n c e a n u m b e r oY word c a n d i d a t e s a r e l i k e l y t o be f o u n d t h r o u~h o u t t h e u t t a r k n e , i t may be f r u i t f u l t o b e a b l e t o s e l e c t a s u b s e t of t h e m o n semantic, p r a g m a t i c , o r p r o s o d i c g r o u n d s a s well a s s y n t a c t i c , d e p e n d i n g o n w h i c h c u e s seem m o s t r o b u s t .", "num": null, "type_str": "figure", "uris": null }, "FIGREF3": { "text": "speech p a r s e r w i l l b e s i m i l a r t o t h e word l a t t i c e d e s c r i b e d a b o v e , t h u s t h e p a r s e r w i l l h a v e t o f a c e n o t o n l y t h e p r o b l e m t h a t o n e o r more w o r d s i n i t s i n p u t m i p h t b e", "num": null, "type_str": "figure", "uris": null }, "FIGREF4": { "text": "5. T h e p a r s e r w i l l h a v e t o h a v e t h e a b i l i t y t o p r e d i c t w o r d s a n d s y n t a c t i c c l a s s e s w h i c h a r e c o n s i s t e n t w i t h p a r t i a l h y p o t h e s e s a b o u t t h e c o n t e n t of t h e s e n t e n c e i n o r d e r t o h e l p f i l l g a p s i nt h e l a t t i c e .6. B e c a u s e of t h e c o m b i n a t o r i a l e x p l o s i o n o f s y n t a c t i c a l t e r n a t i v e s w h i c h o c c u r s when a l l s y n t a c t i c p o s s i b i l i t i e s a r e e x p l o r e d f o r small sections o f a n u t t e r a n c e , t h e s y n t a c t i c component m u s t l i m i t t h e number of s u c h a l t e r n a t i v e s which a r e a c t u a l l y g e n e r a t e d , o r a t l e a s t f a c t o r t h e m o r t r e a t them i m p l i c i t l y r a t h e r t h a n e x p l i c i t l y . One way of p a r t i a l l y s o l v i n g t h i s p r o b l e m is t o o r d e r the a l t e r n a t i v e s i n s u c h a way t h a t o n l y t h e b e s t a l t e r n a t i v e s ape e x t e n d e d . S e c t i o n 2 The BBN S p e e c h U n d e r s t a n d i n 3 S y s t e m I n t h e p a s t few years t h e r e h a s b e e n a f l u r r y of a c t i v i t y i n t h e f i e l d o f a u t o m a t i c s p e e c h u n d e r s t a n d i n g , r e s u l t i n g i n a number o f d i f f e r e n t s y s t e m s . F o r s u r v e y s o f a number of these s y s t e m s t h e reader is recommended t o wolf 1 3 1 1 , B a t e s [ 4 ] , a n d Hyde [ l o ] . Fgr more s p e c i f i c d e t a i l s on soma o f t h e i n d i v i d u a l s y s t e m s , s e a", "num": null, "type_str": "figure", "uris": null }, "FIGREF5": { "text": "I n c . , t h a t s y s t e m i s b r i e f l y d e s c r i b e d h a r e a n d i s f u r t h e r d o c u m e a t a d i n [ 3 , 4 , 5 , 6 , 1 5 , 2 3 , 2 4 , 2 6 , 33 9 351. SPEECHLIS h a s u s e d two t a s k d o m a i n s ; t h a t of t h e LUNAR t e x t q u e s t i o n -a n s w e r i n g s y s t e m 1 3 6 1 w h i c h d e a l s w i t h c h e m i c a l a n a l y s e s o f A p o l l o 1 1 moon r o c k s a n d o n e d e a l i n g w i t h t r a v e l b u d g e t management.", "num": null, "type_str": "figure", "uris": null }, "FIGREF6": { "text": "26]), p h o n o l o g i c a l r u l e s a u g m e n t t h e o u t p u t of t h e a c o u s t i c c o m p o n e n t t o i n c l u d e s e q u e n c e s of p h o n e m e s w h i c h c o u l d h a v e r e a u l t e d i n t h e o b s e r v e d p h o n e m e s ; t h e l e x i c a l r e t r i e v a l c o m p o n a n t r e t r i e v e 8 w o r d s from t h e l e x i c o n o n t h o b a s i s o f t h i s i n f o r m a t i o n ( R o v n e r , e t . a l . [ 2 4 ] ) ; t h e word m a t c h e r determines t h e d r~r e a t o w h i c h t h e i d e a l p h o n e t i c s p e l l i n g of a g i v e n word m a t c h e s t h e a c o u s t i c a n a l y s i s a t a p a r t i c u l a r l o o a t i o n [ 2 4 ] . All of t h e s e c o m p o n e n t s s t r u c t u r e t h e i r o u t p u t i n s u c h a way b s t o r e p r e s e n t t h e a m b i g u i t y whi;*l i s i n h e r e n t i n t h e i r a n a l y s e s . F o r e x a m p l e , t h e y c a n be u s e d t o p r o d u c e w o r d l a t t i c e s s u c h a s t h a t w h i c h was shown i n F i g u r e 1 . 1 . T h e s y n t a c t i c c o m p o n e n t is SPARSER, t h e s y s t e m c o m p r i s i n g t h e b o d y of t h i s p a p e r (see a l s o Bates [ 3 , 4 1 ) . A c c e p t a b l e u t t e r a n c e s are n o t r e s t r i c t e d t o c o n t e x t -f r e e s y n t a x , s i n c e t h e grammar w h i c h SPARSER u s e s i s a r n o q i f i a d ATN grammar, c a p a b l e o f h a n d l i n g a l a r g e , n a t u r a l s u b s e t o f E n g l i s h . The r e m a i n i n g s e c t i o n s of t h i s t h e s i s d e t a i l t h e s t r u c t u r e a n d o p e r a t i o n o f", "num": null, "type_str": "figure", "uris": null }, "FIGREF7": { "text": "a n t i c formalism is v e r y -g e q o r a l a l t h o u g h a new n e t w o r k m u s t be c o n s t r u c t e d f o r e a c h new t a s k d o m a i n . p r a g m a t i c s c o m p o n e n t i s c u r r e n t 1 9 b e i n g i m p l e m e n t e d , b u t i s p r o j e c t e d t o c o n t a i n i n f o r m a t i o n a b o u t t h e p a s t d i a l o g u e , a m o d e l of t h e u s e r , a n d o t h e r p r a g m a t i c d a t a ( s a t ? B r u c e [6]).c o n t r o l c o m p o n e n t c o n t a i n s a n o v e r a l l s t r a t e g e r t o o b t a i n a n i n t e r p r e t a t i o n o f a n u t t e r a n c e (see R o v n s r , e t a1D 3 1 ) . I t d e c i d e s w h i c h c s m p o n e n t i s t o be c a l l e d , w h a t i n p u t i t i s t o begiven, a n d w h a t i s t o b e d o n e w i t h t h e o u t p u t .It s e t s t h r e s h o l d s o n w o r d m a t c h q u a l i t y . I t c o m b i n e s t h e s c o r e s p r o d u c e d by t h e o t h e r c o m p o n e n t s i n o r d e r t o r a n k c o m p e t i", "num": null, "type_str": "figure", "uris": null }, "FIGREF8": { "text": "w i t h a l a r g e ATW .grammar f o r parsing t e x t l a i d t h e grouudwgrk f o r t h e development of a s i m i l a r grammar f o r speech. W e have t r i e d a s much a s p o s s i b l e t o keep the formalism which was developed by Woods i n t a c t , b u t some chances have been necessary o r d e s i r a b l e t o make t h e grammar more amenable t o the speech p a r s e r . W e c a l l the formalism a Modifi& Au~mentdd Transitiom Network ( M A T N ) , and a s s e r t t h a t i t has the same power a s t h e o r i g i n a l ATN formalism. The changes, a r e b r i e f l y indicated h e r e . For a f u l l e r d i s c u s s i o n , s e e Bates [ 4 ] . Every a r c of an ordinary ATN h a s a t e s t component, which may be any p r e d i c a t e . It i s u s u a l l y a boolean combination of tests on the c u r r e n t i n p u t word ( i t s f e a t u r e s , e t c . ) and t h e c o n t e n t s of r e g i s t e r s which have been s a t by a c t i o n s on previous a r c s . I n c u r r e n t word and i t s f e a t u r e s , i . e . a l o c a l , c o n t e x t -f r e e t e s t . The second is a t e s t on t h e register c o n t e n t s , i.e. a c o n t e x t -s e n s i t i v e t e a t . Both t e s t s m u s t succeed f o r t h e a r c t o be taken. The reason f o r s p l i t t i n g u p the t e s t a i n t h i s wav i s t h a t r e g i s t e r checking t e s t s cannot b e made u n l e s s t h e r e p i s t e r s a r e s a t , apd i n mally situations i n t h e speech environment t h e r e may not b e e n o u~h l e f t context t o puarantee t h a t t h e prcper r*egisttbrs would b e s e t . Thua i t is u s e f u l t o be a t o e v a l u a t e t h e c o n t e x t -f r e e t e s t on an a r c a t a d i f f e r e n t time i n t h e parsing Rrocess from t h e c o n t e x t -s e n s i t i v e one. On PUSH a r c s , t h e r e a r e t h e types of t e s t s which a used. I t i s u s e f u l and e f f i c i e n t t o t e s t t h e n e x t word of i n p u t before actually doing t h e PUSH, t o s e a , f o r example, i f t h e next word can begin a c o n s t i t u e n t of t h e t y~e being PUSHed f o r . T h i s t e s t is c a l l e d a look-ahead t e s t , and t a k e s t h e place of t h e normal c o n t e x t -f r e e t e s t i n the test component of t h e a r c . T h e m i s a l s o t h e usual c o n t e x t -s e n s i t i v e test on r e p i s t e r s ;rhich were s e t before t h e PUSH a r c was el-]countered. And f i n a l l y , when t h e PIISH a r c r e t u r n s w i t h a c o n s t i t u e n t , another c o n t e x t -f r e e t e d t may be d o n e on t h e s t r u c t u r e of t h e e n t i r e c o n s t i t u e n t . Therefore, t h e t e s t component of a PUSH a r c is a l i s t o.f the t h r e e t e s t s i u s t d e s a r i b e d . SENDR s were an e f f i c i e n t mechanism f o r t e x t parsing because they allowed t e s t s t o be made on a lower l e v e l w h i c h i n v o l v e d information obtained somewhere ( p o s s i b l y f q r ) t o t h e left i n t h e", "num": null, "type_str": "figure", "uris": null }, "FIGREF9": { "text": "t h e word \" p e r s o n f f i s n o t t h e word t h e p e r s o n who. t r a v e l s . . . , which was r e a l l y u t t e r e d . If i t were a l l o w e d t o b e p a s s e d down", "num": null, "type_str": "figure", "uris": null }, "FIGREF10": { "text": "T h f s i s a p r o c e s s w h i c h would b e e x t r e m e l y w a s t e f u l , e s p e c i a l l y i n t h os p e e c h e n v i r o n m e n t w h e r e one w a n t s t o b e a b l e t o t a k e a s much a d v a n t a g e a3 p o s s i b l e o f i n f o r m a t i o n w h i c h was g a i n e d a t one p o i n t a n d s l i g h t l y a l t e r e d a t a n o t h e r . I n p a r t i c u l a r , i t isa d v a n t a g e o u s t o c o n s i d e r a s c o n s t i t u e n t s s u c h c o n s t r u c t i o n s l -f o r m e d -s u b s t r i n g t a b l e f o r u s e by o t h e r p r o b e s s e s . A n o t h e r r e a s o n i t h a t some t y p d v of v e r i f i c a t i o n s ( s e m a n t i c , p r o s o d i c , a n d p r a g m a t i c , a t l e a s t ) c a n be d o n e m o s t c o n v e n i e n t l y o n p o r t i o n s of a n u t t e r a n c e w h i c h h a v e b e e n a s s i g n e d a s y n t a c t i c s t r u c t u r e , i.e. on c o n s t i t u e n t s . If a p o r t i o n of a n u t t e r a n c e i s p a r s e d (e.g! \" t h a t I g a v e y o u f f f r o m t h e c o m p l e t e u t t e r a n c e \"The book t h a t I g a v e y o u f 1 ) b u t d o a s n o t form a complete c o n s t i t u e n t b e c a u s e i t i s m i s s i n g a p i e c e of i n f o r m a t i o n from a h i g h e r c o n s t i t u e n t t o t h e l e f t w h i c h would h a v e b e a n s e n t down h a d i t been a v a i l a b l e , t h e n t h e s e v e r i f i c a t i o n s may n o t be made u n t i l t h e m i s s i n g word o r w o r d s a r e i d e n t i f i e d . Yet i t may Page 12 be important t o build a n d verify t h e c o n s t i t u e n t in o r d e r t o p r e d i c t t h e missing word to t h o l e f t . Therefore, it i s better t o allow c o n s t i t u e n t s t o be built without information w h i c h would normally have been passed down When parsing p o s s i b l y inoorrect fragments with l i t t l e o r no l e f t c o n t e x t , i t is brat t o keep c o n s t i t u e n t s a3 small a n d a s independent a s p o s s i b l e . The conversion process from an ATN eramrnar t o a MATN prammar with rapard to SENDR s is straiffhtforwardand i n f a l v r s the use of a dummy symbol which is used i n t h e c o n s t r u c t i o n o f t h e lower Level c o n s t i t u e n t . When the s t r u c t u r e i s p o p p e d , t h e PUSH a r c examines i t f o r agreement and may r e p l a c e t h e dummy node by t h e a p p r o p r i a t e item which would have baeb s e n t down. T h e s t r u c t u r e returned b y t h e PUSH f o r a r e l a t i v e c l a u s e on t h e fragment \" t h a t I gave youw might look l i k e F i g u r e 3.1 (where t h d s t r u c t u r e i s shown i n both t h e usual t r e e diagram form a n d a corresponding form more amenable t o computer o u t p u t ) .", "num": null, "type_str": "figure", "uris": null }, "FIGREF11": { "text": "Two Representations of a Parse Tree T h e f o u r t h e l e m e n t o f e v e r y a r c i n a MATN i s a small i n t e g e r w h l c h i s called t h e w e i~h t of t h e a r c . T h i s w e i g h t was o r i g i n a l l y c o n c e i v e d of a s a r o u g h m e a s u r e of e i t h e r ( a ) how l i k e l y t h e a r c i s t o b e t a k e n when t h e p a r s e r i s i n t h a t s t a t e or ( b ) how much i n f o r m a t i o n i s l i k e l y t o be gained from t a k i n g t h i s a r c , i.e. w h e t h e r t h e p a r s e p a t h w i l l b l o c k q u i c k l y i f t h e a r c i s wronn. T h a t t h e s e two s c h e m e s a r e n o t e q u i v a l e n t c a n b e s e e n b y t h e f o l l o w i n g e x a m p l e . I n a g i v e n s t a t e , say j u s t a f t e r t h e maln v e r b of t h e s e n t e n c e h a s bean f o u n d , t h e a r c w h i c h a c c e p t s a p a r t i c l e may be much less l i k e l y t h a n t h e a r c whdch Jumps t o a n o t h e r s t a t e t o l o o k f o r c o m p l e m e n t s . However i f a p a r t i c l e w h i c h agrees w i t h t h e v e r b i s f o u n d i n t h e i n p u t stream a t t h i s p o i n t , t h e n t h e p a r t i c l e a r c i s more l i k e l y t o b e c o r r e c t . S i n c e it is n o t a t a l l c l e a r how t o m e a s u r e o r e v e n i n b u i t how much i n f o r r n a t i o p is l i k e l y t o be g a i n e d from t a k i n g a n a r c , i t was d e c i d e d t h a t t h e w e i g h t s would r e f l e c t r e l a t i v e l i k e l i h o o d s . T h e a c t u a l w e i g h t s w h i c h h a v e b e e n u s e d i n t h e s p e e c h grammar r e f l e c t a n T n t u i t i v e , t h o u g h e x p e r i e n c e d g u e s s a s t o how l i k e l y t h e a r c i s t o be c o r r e c t i f i t is t a k e n , a s s u m i n g the s t a t e i t s e l f is on t h e c o r r 2 c t p a t h .", "num": null, "type_str": "figure", "uris": null }, "FIGREF12": { "text": "d e c l a r a t i v e s , noun p h r a s e u t t e r a n c e s , i m p e r a t i v e s , a c t i v e and p a s s i v e forms, r e l a t i v e clauses ( r e d u c e d a n d u n r e d u c a d ) , c o m p l e m e n t s , s i m p l e q u a n t i f i e r s , noun-noun m o d i f i e r s , v a r b -p a r t i c l e c o n s t r u c t i o n s , n u m b e r s , a n d d a t e s ( b u t n o t c o n j u n c t i o n s ) . I t b e g a n as a m o d i f i c a t i o n of t h e grammar for t h e Page 14 LUNAR system [361 but has been considerably adapted and expanded. T h i s grammar is. called SPEECHGRAMMAR, and is l i s t e d ina[4].", "num": null, "type_str": "figure", "uris": null }, "FIGREF13": { "text": "probably shows t h e structure mor-c c l e a r l v . The s e r i o u s r r o d r r is encouraged t o sketch n ccpy of t h i s grammar f o r r-rfrrencc l a t e r e work reported here was f i n i s h e d , t h e author has written another grammar, called SMALLGRAM which uses t h e 1IATN formalism but which embodies a great deal of semantic and pragmatic, information s p e c i f i c to t h e domain of discourse currently baing used by t h e BBN speech understand in^ project.", "num": null, "type_str": "figure", "uris": null }, "FIGREF14": { "text": "i n c e t h e ramm mar i s o r g a n i z e d f o r n o r m a l p a r s i n g i n j u s t t h e o p p o e i t e f a s h i o n , i . e .", "num": null, "type_str": "figure", "uris": null }, "FIGREF15": { "text": "i t was n e c e s s a r y t o b u i l d an i n d e x i n t o the granlmar. T h i s i n d e x consists of a number o f t a b l e s c e n t a i n i n g pre-computed i n f o r m a t i o n w h i c h i n e f f e c t i n v e r t s t h e grammar. S e c t i o n 4 U v e r v i e w of SPARSER The i n p u t t o SPARSER i s assumed t o b e a s e t of w o r d s together w i t h t h e i r b o u n d a r y p o i n t s ( w h i c h may o r may n o t be r e l a t e d to p o i n t s in time). A word together with its boundaries I s termed a word m a t c h . A word match a l s o i n c l u d e s a score which indicates how w e l l t h e i d e a l p h o n e m i c r e p r e s e n t a t i o n of t h e word matched the acoustic a n a l y s i b of t h e u t t e r a n c e ( b u t as w e s h a l l see the parser h a s l i t t l e n e e d o f t h i s i n f o r m a t i o n ) .", "num": null, "type_str": "figure", "uris": null }, "FIGREF16": { "text": "same word may m a t c h a t several sets o f b o u n d a r y p o i n t s o r may match in deveral ways b e t w e e n t h e same b o u n d a r y p o i n t s , each w o r d ~t~~ is also g i v e n a u n i q u e number t o h e l p i d e n t i f y i t . Thus t h e s t r u c t u r e f o r a basic w o r d m a t c h i s : (number word l e f t b o u n d a r y r i g h t b o u n d a r y l e x i c a l s c o r e ) 5 11 9 4 ) , o r ( 4 TRAVEL 5 11 (94 1 1 0 ) ) w h e r e t h e score i s g i v e n a s a p a i r of n u m b e r s r e p r e s e n t i n g t h e a c t u a l a n d maximum s c o r e s , o r (4 TRAVEL 5 1 1 ) w h e r e t h e s c o r e i s o m i t t e d . How i s t h e i n p u t t o t h e p a r s e r t o be c o n s t r u c t e d ? W e a s s u m e t h a t a c o u s t i c p r o c e s s i n g a n d l e x i c a l s c a n n i n g c o m p o n e n t s c a n o p e r a t e on a d i g i t i z e d waveform t o p r o d u c e a number o f word m a t c h e s s u c h a s p r e v . i o u s l y shown i n t h e word l a t t i c e o f F i g u r e 1 . 1 . ( T h a t t h i s i s p o s s i b l e h a s b e a n d e m o n s t r a t e d by Woods [ 3 3 ] ) . A l l o w i n g t h e p a r s e r t o o p e r a t e u n r e s t r i c t e d on t h e e n t i r e word l a t t i c e would p r o b a b l y n o t b e f r u i t f u l b e c a u s e o f t h e l a r g e n u m b e~ o f l o c a l l y s y n t a c t i c a l l y c o r r e c t c o m b i n a t i o n s o f w o r d s , b u t one p o s s i b i l i t y f o r i n p u t to t h e p a r s e r w o u l d be t o t a k e a s e t o f t h e b e s t -m a t c h i n g , n o n -o v e r l a p p i n g word m a t c h e s i n t h e l a t t i c e , s u c h a s t h o s e i n F i g u r e 4 . 1 . A s e t o f n o n -o v e r l a p p i n g word matches i s a h y p o t h e s i s a b o u t t h e c o n t e n t o f t h e u t t e r a n c e . I n o r d e r t o a v o i d c r e a t i n g l a r g e n u m b e r s o f s u c h s e t s which a r e p u t t o g e t h e r c o m b i n a t o r i a l l y w i t h n o b a s i s e x c e p t l o c a l a c o u s t i c m a t c h , s e m a n t i c o r p r a g m a t i c processes c a n b e u s e d t o g r o u p w o r d matches b a s e d on w h a t i s m e a n i n g f u l o r l i k e l y t o be h e a r d . F o r e x a m p l e , i f a d i a l o g u e h a s been a b o u t v a r i o u s n i c k e l compounds, t h e c o m b i n a t i o n \" n i c k e l a n a l y s e s w may b e more l i k e l y t h a n \" c h e m i c a l a n a l y s e s f f e v e n t h o u g h t h e word match f o r ' t c h e m i c a l f f h a s a h i g h e r s c o r e t h a n t h a t f o r m n i c k e l f f f ' . W e w i l l n o t attempt t o d e t a i l here how t h i s s e m a n t i c g r o u p i n g c o u l d . b e d o n e a n d how t h e sets c o u l d be s c o r e d , s i n c e i t h a s b e e n d e s c r i b e d e l s e w h e r e [ 1 5 ] .", "num": null, "type_str": "figure", "uris": null }, "FIGREF17": { "text": "Sample Word Match S e t s U s i n g more t e r m i n o l o g y from t h e BBN s p e e c h s y s t e m , t h e word t h e o r v t o d e n o t e s a set of word m a t c h e s s u c h a s we h a v e j u s t d e s c r i b e d t o g e t h e r w i t h ( p o s s i b l y e m p t y ) s l o t s f o r i n f o r m a t i o n f r o m e a c h o f t h e p o s s i b l e k n o w l e d g e s o u r c e s i n t h e s y s t e m . From t h e p o i n t o f v i e w . o f SPARSER, u s u a l l y o n l y t h e word match p o r t i o n o f a t h e o r y i s of f n t e r e s t , h e n c e we s h a l l f a l l i n t o t h e h a b i t o f u s i n g t h e word \" t h e o r y v t o r e f e r t o t h e word match s e t i t c o n t a i n s . When s p e a k i n g o f t h e s y n t a c t i c component of a t h e o r y , h o w e v e r , w e a r e r e f e r i n g t o t h e i n f o r m a t i o n s l o t f o r s y n t a x whicn a c c o m p a n i e s each word match s e t . T h e o r i e s h a v e t h e f a l l o w i n g c h a r a c t e r i s t i c s : 1 ) They c o n t a i n a s e t o f b a s i c , n o n d v e r l a p p i n g word m a t c h e s . 2 ) They t e n d a t f i r s t t o c o n t a i n l o n g c o n t e n t w o r d s a n d n o t many s h d r t f u n c t i o n w o r d s . T h i s i s b e c a u s e l o n g w o r d s a r e more r e l i a b l y a c o u s t i c a l l y v e r i f i e d a n d c o n t e n t w o r d s a r e e a s i e r t o P a g e 18 r e l a t e s e m a n t i c a l l y a n d p r a g m a t i c a l l y . S i n c e small w o r d s s u c h a s t a m r~d~v , r r t h e r l , r r~n a l l , \" h a v e \" , rr Of 11 11 i n rr I T , e t c . may b e r e p r s s e n t e d b y v e r y l i t t l e a c a Q s t i c i n f o r m a t i o n , t h e y w o u l d t e n d t o m a t c h a t many p l a c e s i n t h e u t t e r a n c e w h e r e t h e y d o n o t r e a l l y o c c u r . C o n s e q u e n t l y t h e y ase n o t s e a r c h e d for? b y t h e i n i t i a l w o r d m a t c h s c a n , n o r a r e t h e y p r o p o s e d i n t h e s e m a n t i c s t a g e s of h y p o t h e s i s f o r m a t i o n .", "num": null, "type_str": "figure", "uris": null }, "FIGREF18": { "text": "y d o n o t ) c o m p l e t e l y s p a n t h e u t t e r a n c e , b u t have n u m e r o u s g a p s of v a~v i n g sizes ( a . p . f o r t h e f u n c t i o n w o r d s ) . 4 ) They t e n d t o c o n t a i n some s e q u e n c e s of c o n t i g u o u s w o r d m a t c h e s . S u c h a s e q u e n c e is c a l l e d a n i s l a n d . T h a t s u c h a s e t of t h e o r i e s c a n b e c r e a t e d h a s b e p o n e n t , SPARSER, is e x p e c t e d t o p r o c e s s t h e s e t h e o r i e s one a t a time. I n c e r t a i n c i r c u m s t a n c e s w h i c h w i l l b e d e t a i l e d l a t e r , t h e i n p u t t o SPARSER w i l l b e a t h e o r y t o g e t h e r w i t h o n e o r more word m a t c h e s w h i c h a r e t o be a d d e d i n o r d e r t o c r e a t e a new l a r g e r t h e o r y w h i c h i s t h e n t o be s y n t a c t i c a l l y a n a l y z e d . W e w i l l a s s u m e t h a t t h e r e e x i s t s a c a n t r o l c o m p o n e n t w h i c h p r e s e n t s SPARSER w i t h t h e o r i e s t o p r o c e s s a n d t o w h i c h SPARSER c a n c o m m u n i c a t e p r e d i c t i o n s a n d r e s u l t s . P r e l i m i n a r i e s G i v e n a t h e o r y , w h a t i s t o be d o n e w i t h i t ? W e b e g i n by c o n s i d e r i n g a s u b s e t of t h e q u e s t i o n : G i v e n a n i s l a n d o f w o r d m a t c h e s , w h a t is t o b e d o n e w i t h i t ? T h e a n s w e r is t o c r e a t e o n e P a g e 19 o r more parse p a t b n t h r o u g h t h o i s l a n d a n d t o p r e d i c t w h a t words o r s y n t a c t i c c l a s s e s c o u l d s u r r o u n d t h e i s l a n d . A p a r s e p a t h i s t h o S e q u e h c e o f a r c s i n t h e grammar w h i c h w o u l d b e u s a d b y a c o n v e n t i o n a l ATN p a r s e r t o p r o c e s s t h e w o r d s i n t h e i s l a n d , i f t h e i s l a n d were embedded i n a c o m p l e t e s e n t e n c e . For e x a m p l e , c o n s i d e r t h e way a p a r s e r m i g h t p r o c e s s N A L Y S E S 22 30) u s i n g t h e M I N I G R A M M A R of t h e p r e v i o u s s e c t i o n .", "num": null, "type_str": "figure", "uris": null }, "FIGREF19": { "text": "( o m i t t i n g f o r t h e moment t h e p r o b l e m of how i t is known t h a t NP/ i s t h e r i e h t p l a c e t o b e g i n ) t h e s e q u e n c e of s a r c s w h i c h w o u l d bh t a k e n t o p a r s e \" c h e m i c a l analysesw a s a noun p h r a s e i s t h a t shown b e l o w i n t i o n o f M I N I G R A M M A R n e e d e d t o p a r s e % h e m i c a 1 a n a l y s e s m Let u s d e f i n e a c o n f i a t m a t i o n t o b e a r e p r e s e n t a t i o n o f t h e", "num": null, "type_str": "figure", "uris": null }, "FIGREF20": { "text": "f i g u r a t i o n r e p r e s e n t s a s t a t e w h i c h i s e i t h e r t h e i n i t i a l s t a t e of t h e grammar o r a s t a k e w h i c h c a n be PUSHed t o ( i . e a 8rl i n i t i a h c o n f i g u r a t i o n , a n d is i n d i c a t e d s c h e m a t i c a l l y by a f i l l e d -i d semi-circle a t t a c h e d t o t h e l e f t edge of t h e b o x . N o t e t h a t a c o n f i~u r a t i o n N~/ Q U A N T : I~ i s q u i t e d i s t i n c t f r o m a c o n f i g u r a t i o n NP/QUANT:22 s i n c e t h e y a r e a t d i f f e r e n t p o s i t i o n s i n t h e i n p u t . I n SPARSER, each c o n f i g u r a t i o n .is a l s o a s s i g n e d a u n i q u e number w h i c h i s a c o n v e n i e n t i n t e~n a l p o i n t e r . The p r o c e s s of t r a v e r s i n~ a n a r c o f t h e grammar u s i n g a p a r t i c u l a r word i s r e p r e s e n t e d b y a t r a n s i t i o n from one c o n f i g . u r a t i o n t o a n o t h e r . A t r a n s i t i o n e a n be made o n l y i f t h e a r c t y p e i s c o m p a t i b l e w i t h t h e c u r r e n t item o f i n p u t a n d i f t h t e x t -s e n s i t i v e tests a r e e v a l u a t e d l a t e r . ) A t r a n s i t i o n c a r r i e s w i t h i t i n f o r m a t i o n a b o u t t h e a r c w h i c h i t r e p r e s e n t s a n d t h e item o f i n p u t i t u s e s . T h e item o f i n p u t i s u s u a l l y t h e word m a t c h w h i c h t h e a r c u s e s , b u t i t is N I L i n cases s u c h as JUMF a r c s w h i c h d o n o t u s e i n p u t , a n d i t is a c o m p l e t e c o n s t i t u e n t f o p PUSH a m s . A u n i q u e i d e n t i f y i n g number a n d t h e l i s t o f f e a t u r e s , i f a n y , w h i c h is a s s o c i a t e d w i t h t h e i n p u t word o r c o n s t i t u e n t are el30 recorded o n t h e t r a n s i t i o n i n SPARSER, b u t t h e y a r e n o t shown s c h e m a t i c a l l y . A t r a n s i t i o n i s r e p r e s e n t e d s c h e m a t i c a l l y by a n arrow from o n e c o n f i g u r a t i o n t o a~~o t h e r w i t h an a b b r e v i a t e d f o r m of t h e a r c w r i t t e n a b o v e t h e arrow a n d t h e item o f i n p u t u n d e r i t . T h e s y n t a c t i c p a r t of a n y t h e o r y w h i c h SPARSER processes c o n t a i n s , among o t h e r t h i n g s , l i s t s o f t h e t r a n s i t i o n s a n d c o n f i g u r a t i o n s w h i c h a r e created o r u s e d b y t h e t h e o r y . T h u s wheh w e t a l k a b o u t c r e a t i n g a c o n f i g u r a t i o n o r t r a n s i t i o n i t i s i m p l i c i t l y understood t h a t SPARSER a l s o adds i t t o the", "num": null, "type_str": "figure", "uris": null }, "FIGREF21": { "text": "a complete p a t h , o t h e r w i s e i t i s a p a r t i a L p a t h . Paths a r e assumed t o be p a r t i a l u n l e s s otherwise s p e c i f i e d . darrinnina Paraa a I s l a n d SPARSER p r o c e s s e s an i s l a n d of words by beginning w i t h t h e l e f t m o s t word and determining i t s p o s s i b l e p a r t s of speech. Then t h e a r c s of t h e grammar which can process t h e word a r c f p u n d ( b y looking i n the previoRdly c o n s t r u c t e d Rrammar i n d e x ) . For each a r c , two c o n f i~u r a t i o n s a r e c o n s t r u c t e d one f o r t h e s t a t e a t t h e t a i l of t h e a r c and one f o r t h e s t a t e a t t h e head, u s i n g t h $ l e f t and r i g h t boundary p o s i t i o n s of t h e word match, r e s p e c t i v e l y , and a t r a n s i t i o n f o r t h a t a r c u s i n g t h e c u r r e n t word match i s a l s o b u i l t . Schematically, we have f o r our example a s i t u a t i o n which looks l i k e t h a t of Figure 4 . 4 ( s u c h a d i s p l a y of a l l o r some of t h e t r a n s i t i o n s and c o~f i . q u l s t r u c t e d i s c a l l e d a m a p ) . Notice t h a t a c o n f i g u r a t i o n mayhave any number of t r a n s i t i o n s e n t e r i n g o r l e a v i n g i t .", "num": null, "type_str": "figure", "uris": null }, "FIGREF22": { "text": "n i t i a l map f o r p a r s i n g llchemical a n a l y s e s t r The idea of t h i s process i s t o begin t6 s e t up paths which my be used t o parse t h e i s l a n d . However i t is not n e c e s s a r i l y t h e case t h a t the only donfigurations which could s t a r t paths t h r o u~h the island a r e those which have just been obtained, s i n c e i t may be possible t o o r e a t r t r a n s i t i o n s which e n t e r them via JUMP a r c s o r TST ahca.For each s t a t e , t h e sequence of a r c s which can reach i t without u s i n g t h e previous word of i n p u t have", "num": null, "type_str": "figure", "uris": null }, "FIGREF23": { "text": "Note t h a t any of t h e configurations (except f o r NP/ADJ:22 22 J J * and NP/N:22) could a c t u a l l y be t h e c o r r e c t l e f t m o s t configuration f o r t h i s i s l a n d , depending upon what t h e ( c u r r e n t l y unknown) l e f t n t h e grammar index, SPARSER can determine, f o r t~c each configuration which could s t a r t t h e i s l a n d , j u s t what s o r t", "num": null, "type_str": "figure", "uris": null }, "FIGREF24": { "text": "a d j e c t i v e could precede t h e i s l a n d and, i f i t d i d , t h e t r a~s i t i o n w h i c h would proceas i t would t e r m i n a t e on c o n f i g u r a t i o n NP/ADJ:14, Because the i n i t i a l c o n f i g u r a t i o n NP/:14 could s t a r t t h e i s i a n d , anything which could precede n noun phrase could occur t o t h e l e f t ; again t h e grammar i n d e x provides t h e information t h a t the CAT PREP a r c could l e a d t o a c o n f i g u r a t i o n w h i c h could accept a noun phrase ( v i a t h e PUSH NP/ a r c ) , so a p r e p o s i t i o n could also p r e f i x t h e i s l a n d . If t h e index f u n c t i o n s i n d i c a t e t h a t a c o n s t i t u e n t could be picked u p by a PUSH a r c which could t e r m i n a t e on t h e c o n f i g u r a t i o n under c o n s i d e r a t i o n , an i n d i c a t i o n i s made i n tho WFST s o t h a t any time a c o n s t i t u e n t of t h e d e s i r e d type i s b u i l t which ends a t t h e proper l o c a t i o n , i t may be t r i e d h e r e . Because of t h e highly r e c u r s i v e n a t u r e of ATN grammars, i t i s vary l i k e l y t h a t a s we chain back through t h e p o s s i b l e sequences of PUSHas which could lead t o tho beginning of tho c u r r e n t c o n s t i t u e n t ( o r t h e seauence of POPS w h i c h could be i n i t i a t e d by t h e completion of t h e c u r r e n t c o n s t i t u e n t ) a l a r g e number of p r e d i c t i o n s w i l l be made. Rather than make a l l t h e s e p r e d i c t i o n s a u t o m a t i c a l l y , beford we a r e even s u r e t h a t t h e r e i s i n f a c t a c o n s t i t u e n t a t t h e c u r r e n t l e v e l , t h e p o s s i b l e c o n f i g u r a t i o n s which could make p r e d i c t i o n s on o t h e r l e v e l s a r e saved t o be a c t i v a t e d l a t e r i f t h e p r e d i c t i o n s from t h e c u r r e n t s e t of a c t i v e c o n f i g u r a t i o n s a r e not s u f f i c i e n t . T h e p r e d i c t i o n s w h i c h a r e made ( n o t s a v e d ) are n o t acted upon a t t h i s time, b u t a r d k e p t i n t e r n a l l y b y SPARSER u n t i l a l l t h e i s l a n d s o f t h e t h e o r y h a v e b e e n p r o o e s s e d . W e s h a l l see b d e w w h a t t h e n b e c o m e s of t h e p r e d i c t i o n s . I s l a n d Once p r o c e a s i n g h a s p r o c e e d e d t h i s f a r , we c a n go b a c k a n d c o n s i d e r t h e s e t of c o n f i g u r a t i o n s w h i c h r e p r e s e n t s t a t e s t h e p a r s e r c o u l d be i n just a f t e r p r o c e s s i n g t h e f i r s t word of t h e i s l a n d . In o u r e x a m p l e , these a r e c o n f i g w a t i q n s WP/ADJ:22 a n d NP/N:22. C o n f i g u r a t i o n s s u c h a s t h o s e w h l c h a r e w a i t i n g t o be e x t e n d e d t o t h e r i g h t a r e c a l l e d a c t i v e c o n f i g u r a t i o n s . SPARSER s e l e c t s a s u b s e t of t h e s e t of a c t i v e c o ' n f i~u r a t i o n s (how t h i s s u b s e t i s s e l e c t e d w i l l be d i s c u s s e d i n t h e n e x t s e c t i o n ) a n d f o r e a c h c o n f i g u r a t i o n tries t o e x t e n d i t b y t r y i n 8 t o parse t h e rest o f t h e island b e g i n n i n g i n t h a t c o n f i p u r a t i o n . When t h e parser i s c o n s i d e r i n~ a c o n f i g u r a t i o n a t some p o s i t i o n , t h e i n p u tp o i n t e r is s e t t o t h e word m a t c h of t h e i s l a n d , i f a n y , w h i c h begins a h t h e same p o s i t i o n i n t h e i n p u t .", "num": null, "type_str": "figure", "uris": null }, "FIGREF25": { "text": "t h e p a t h . W e w i l l c o n s i d e r e a c h t y p e of a r c i n t u r n , s i n c e t h e e f f e c t s of t a k i n g v a r i o u s t y p e s of arca a r e d i f f e r e n t , a n d e x p l a i n f o r each case w h a t h a p p e n s i f t h e arc i s t a k e n . W h e t h e r j u s t o n e t r a n s i t i o n , o r s e v e r a l , o r a l l p o s s i b l e t r a n s i t i o n s a r e made from a n a c t i v e c o n f , i g u r a t i o n i s a matter t o be d i s c u s s e d i n S e c t i o n Five.Soma JUMP a r c s d o n o t l o o k a t t h e c u r r e n t item, s o t h e y may be t a k e n w h e t h e r t h e i n p u t p o i n t e r i s s e t t o a word m a t c h o r t o N I L . The t r a n s i t i o n w h i c h r e s u l t s from taking an a r c of t h i s t y p e has a n u l l item a s s o c i a t e d w i t h i t , even if there is a word m a t c h i n t h e t h e o r y a t t h i s p o i n t .Th e p o s i t i o n s of t h e c o n f i p u r a t i o n s a t e a c h e n d of t h e t r a n s i t i o n are t h e same; t h i s c o r r e s p o n d s t o t h e f a c t t h a t an A T N p a r s e r w o u l d n o t move t h e i n p u t p o i n t e r a s a , c o n s e q u e n c e of t a k i n g t h i s a r c . R a r e l y , a JUMP a r c may t e s t t h e c u r r e n t i t e n i n some way, f o r e x a m p l e , t o make a feature check. If t h e r e i s n o word m a t c h f o r i n p u t , an a r c of t h i s t y p e c a n n o t b e t a k e n . If t h e r e i s a word m a t c h , i t is n o t e d o n t h e t r a h s i t i o n wh c h i s c r e a t e d , b u t t h e c o n f i g u r a t i o n s a t e a c h e n d of the t r a n s i t i o n h a v e t h e same p o s i t i o n . ( I t i s t h a n t h e case t h a t t h a i n e x t i n p u t -u s i n g o r i n p u t -c o n s u m i n g t r a n s i t i o n o n t h e p a t h i n c l u d i n g t h i s t r a n s i t i o n must u s e t h e same word m a t c h . ) T h e s e are TST, CAT, a n d WRD a r c s w h i n e x t s t a t e ) a c t i o n . The o p e r a t i o n i s e x a c t l y t h o same a s t h a t a b o v e e x c e p t t h a t the c o n f i g u r a t i o n o n w h i c h t h e t r a n s i t i o n t e r m i n a t e s h a s t h e p o s i t i o n of t h e r i g h t b o u n d a r y of the c u r r e n t w o r d m a t c h . T a k i n g a POP a r c r e s u l t s i n t h e c r e a t i o n of a t r a n s i t i o n w h i c h h a s a n u l l f i n a l c o n f i g u r a t i o n a n d a n u l l item, b e c a u s e POP a r c s a r e n o t p e r m i t e d t o l o o k a t i n p u t . When a PUSH a r c i s e n c o u n t e r e d , a m o n i t o r i s p l a c e d i n t h e Well-Formed S u b s t r i n g T a b l e (WFST) a t t h e c u r r e n t D o s i t i o n t o await t h e o c c u r r e n c e o f a c o n s t i t u e n t of t h e r e q u i r e d t y p e . If one o r nore s u c h c o n s t i t u e n t s a r e a l r e a d y i n t h e t a b l e , t h e n f o r e a c h o n e t h e r e a r e t h r e e p o s s i b i l i t i e s : i t may b e composed of word m a t c h e s w h i c h a r e i n t h e c u r r e n t t h e o r y , i t may b e composed of word m a t c h e s some of w h i c h a r e n o t i n t h e c u r r e n t t h e o r y b u t which c o u l d be added w i t h o u t v i o l a t i n g t h e n o n -o v e r l a p p i n g c o n s t r a i n t , o r i t may b e composed of word m a t c h e s some o f w h i c h a r e i n c o m p a t i b l d w i t h t h e c u r r e n t t h e o r y . I n t h e f i r s t caae a t r a n s i t i o n i s s e t u p u s i n g t h e c o n s t i b u e n t a s t h e c u r r e n t w o r d . T h e t r a n s i t i o n t e r m i n a t e s on a c o n f l e u r a t i o n whose s t a t e i s d e t e r m i n e d from t h e t e r m i n a t i o n of --. t h e PUSH a r c a n d w h o s e p o s i t i o n is t h a t o f t h e r i g h t b o u n d a r y of t h e r i g h t m o s t word m a t c h i n t h e c o n s t i t u e n t . I n t h e s e c o n d case, a n o t i c e i s c r e a t e d a n d s e n t t o t h e control c o m p o n e n t . A n o t i c e is a r e q u e s t t h a t SPARSER be c a l l e d t o enlarge a t h e o r y by a d d i n p some new i n f o r m a t i o n , i n t h i s c a s e , some a d d i t i o n a l word m a t c h e s w h i c h form a c o n s t i t u e n t t h a t t h e theory c a n u s e . SPARSER d o e s n o t t r y t o d e t e r m i n e when (or e v e n w h e t h e r ) t h e t h e o r y s h o u l d be s o e n l a r g e d . T h a t i s a n i s s u e f o r t h e m a i n c q n t r o l l e r t o d e c i d e ( s e e R o v n e r , e t . a l . l d i s c u s s below how SPARSER e n l a r g e s a t h e o r y i f c a l l e d upon t o d o s o . I n t h e f i n a l c a s e , i f t h e r e a r e n o u s a b l e c o h s t i t u e n t s i n the WFST, a new c o n f i g u r a t i o n i s s e t u p t o s t a r t l o o k i n g f o r o n e a n d i s a d d e d t o t h e l i s t of a c t i v e c o n f i g u r a t i o n s . Its s t a t e i s P a g e 28 t h e s t a t e s p e c i f i e d by t h e PUSH a r c a n d i t s p o s i t i o n i s t h e same a s t h e c u r r e n t c o n f i g u r a t i o n . T h e r e i s a c o n s i d e r a b l e a m o u n t of p r o c e s s i n c t h a t can h a p p e n a n y time o n e of t h e t r a n s i t i o n s l u s t discussed i s rnndr. Whenever% a n i n i t i a l c o n f i g u r a t i o n is c o n s t r u c t e d , t h i s f a c t is r*t)c@rded i n t h e c o n f i g u r a t i o n . Whenever a t r a n s i t i o n i s nade from such a c o n f i p u r a t i o n , t h e i n f o r m a t i o n t h a t t h e r e i s a p a t h I sene initial conf igur-at i o n is recorded on t h e subsoquenf c o n f i g u r a t i o n . S i m i l a r l y , w h e n e v e r a POP t r -a n s i t i o n i s made, t h e c~n~f i g u r a t i o n i t emanates from a n d all p r e v i o t l s c o n f i g u r a t i m s on a n y p a t h which can t e r m i n a t e w i t h t h e POF transition a r e m a r k e d t o i n d i c a t e t h a t t h e y c a n r e a c h a POP t r a n s i t i o n . W h e n e v e r a t r a n s i t i o n i s made w h i c h c o m p l e t e s a p a t h from a n i n i t i a l c o n f i g u r a t i o n t o a POP t r a n s i t i o n , t h e p a t h i s e x e c u t e d , o n e t r a n s i t i o n a t a time, a n d t h e r q i s t e r s e t t i n g a c t i o n s and c o n t e x t s e n s i t i v e tests a r e e x e c u t e d . If a t e s t fails o r an arc a b o r t s , t h e t r a n s i t i o n s a n d configurations f t h e p a t h are removed from t h e l i s t of c o n f i g u r a t i~n s a n d t r a n s i t i o n s w h i c h a r e i n the s y n t a c t i c p a r t of t h e c u r r a n t t h a o r y ( u n l e s s they are u s e d b y a n o t h e r p a t h i n t h e t h e o r y ) b u t n o t removed f r o m t h e n a p . If t h e e x e c u t i o n i s successful, a d e e p s t r u c t u r e t r e e i s p r o d u c e d .", "num": null, "type_str": "figure", "uris": null }, "FIGREF26": { "text": "T h a t s t r u c t u r e t o g e t h e r w i t h i t s f e a t u r e s i s g i v e n a s c o r e , w h i o h may i n c l u d e e v a l u a t i o n s by o t h e r c o r n p o d a n t s s u c h a s s e m a n t i c s a n d prosodies, a n d i s e n t e r e d i n t h e WFST.I t i s q u i t e i m p o r t a n t t h a t s o u r c e sof k n o w l e d g e o t h e r t h a n s y n t a x b e c a l l e d upon t o v e r i f y a n d t o r a n k s y n t a c t i c c o n s t i t u e n t s . T h i s i s b e c a u s e t h e r e a r e l i k e l y t o be many Page 29 ombinations of plausible words from the word lattice which form ayntact%rally reasonable constituents but which may be ruled out om @%her grounds. To allow immediate use of this information which syntax cannot provide alone, SPARSER has an interface to h e semantic component so that constituents can be vrrif ied d i n c t l g without going through the control component. It will be %trivial modification to insert verification aalls t o pragmatics aad prosodies when they become available. In the meantime, even semantic knowledge can be turned off; if the parser gets no IngosnratIon from the call to semantics, it proceeds without it. Plaoement of a constituent in the WFST causes a number of Lbiags to happen. First, any monitors which have been set by the mmnt theory at that position aFe activated. That is, for each ebnfigaration which was waiting for this constituent, a PUSH tmaaitioa 3s made which uses the constituent as its input item. If ao rmonltors have been set which can use this constituent, it as treated exactly as if it were the first word of an island: a11 the PUSH arcs which can use it are found in the grammar index m d appropriate configurations and transitions (including lead-in t n o s i t i o n s , if appropriate) are set up. Next, if there are any monitors for other theories which can use the constituent, patfees are created and output to Control as was described above is the section on PUSH transitiohs.", "num": null, "type_str": "figure", "uris": null }, "FIGREF27": { "text": "6 shows SPARSER s map after our example island has k e n completely processed. The parsing results-in the creation oi a CAT II t r a n s i t i o~ to configuration NP/N:30 using the word *~~a l y s e s~ The PUSH PP/ arc at state NP/N would oause configuration PP/:30 t o be created. Similarly, PP/:22 would be created when t h o configuration NP/N:22 i s picked u p t o be extended.The POP arc transitions from each of the configurations for s t a t e NP/N result in the formation of complete paths, resulting i n the creation of two noun phrases (\"chemical analysesw and ttchamicalft). Since there were no monitors for them, t h e y r e s u l t i n the creation of configuration PP/PREP:14 and $8 subsequent paths.", "num": null, "type_str": "figure", "uris": null }, "FIGREF28": { "text": "i n q an I s l a n d I t may be t h e case t h a t n o p a t h c a n be f o u n d from o n e e n d o f a n i s l a nd t o t h e o t h e r , ( T h i s would o c c u r when a l l a c t i v e c o n f i g u r a t i o n s b l o c k . ) I n t h i s case, t h e r e i s n o p o s s i b l e way t h a t t h e i s l a n d c o u l d form p a r t of a grammatical s t r i n g , s o SPARSER c a n i n f o r m t h e c o n t r o l c o m p o n e n t t h a t t h e t h e o r y i s w r o n g , When a n a c t i v e c o n f i g u r a t i o n i s p i c k e d u p t o b e e x t e n d e d a n d t h e r e i s n o word m a t c h a t t h a t p o i n t , t h e e n d of t h e i s l a n d h a s b e e n r e a c h e d . T h a t d o e s n o t mean t h a t n o more t r a n s i t i o n s c a n b e made, s i n c e a r c s w h i c h d o n o t t e s t t h e i n p u t word c a n be t a k e n a s u s u a l . Arcs w h i c h d o u s e i n p u t c a n n o t b e t a k e n , b u t t h e y c a n b e u s e d t o p r e d i c t w h a t s o r t o f i n p u t would b e a c c e p t a b l e a t t h a t p o s i t i o n . F o r e x a m p l e , a GAT V a r c w h i c h h a s a t e s t r e q u i r i n g t h e v e r b t o be u n t a n s e d w o u l d a l l o w SPARSER t o p r e d i c t a n u n t e n s e d v e r b b e g i n n i n g a t t h e p o s i t i o n of t h e o u r r e n t c o n f i g u r a t i o n . CAT a n d WRD a r c s c a u s e t h e p r e d i c t i o n o f s y n t a c t i c c a t e g o r i e s a n d s p e c i f i c w o r d s , r e s p e c t i v e l y , m o d i f i e d by t h e C o n t e x t -f r e e t e s t on t h e a r c . TST arcs p r o v i d e o n l y t h e t e s t w h i c h m u s t be s a t i s f i e d , a n d PUSH a r c s c a u s e a m o n i t o r t o be s e t i n t h e WFST as w e l l as a TST m o n i t o r f o r t h e t h e l o o k -a h e a d t e s t ( i f a n y ) on the a r c . B n d i n q 3 T h e o r v When k11 t h e i s l a n d s of a t h e o r y h a v e b e e n p r o c e s s e d i n t h e m a n n e r j u s t d e s c r i b e d , i t i s time t o d e a l w i t h t h e g a p s b e t w e e n the i s l a n d s . A s we h a v e seen, arcs i n t h e grammar w h i c h c a nPage 32 e n t e r c o n f i g u r a t i o n s a t t h e l e f t end of an i s l a n d o r which can l e a v e c o n f i g u r a t i o n s a t t h e r i g h t and of an i s l a n d can be used t o make p r e d i c t i o n s about warda t h a t may be a d j a c e n t t o t h e i s l a n d .The p r e d i c t i o n i s a l i s t of t h e a r c , t h e c o n f i~u r a t i o n i t would connect t o , a n d an i n d i c a t i o n of whether t h e t r a n s i t i o n caused by t h e a r c w i l l e n t e r t h e c o n f i g u r a t i o n from t h e l e f t o r leave i t t o t h e r i g h t . If a gap between two i s l a n d 8 i s small e n o u~h t h a t i t may c o n t a i n j u s t one word, than i t is l i k e l y t h a t tho a r c which would process t h a t word may have caused a p r e d i c t i o n from both t h o l e f t and r i g h t s i d e s of t h e gap. If t h i s i s t h e c a s e , and i f t h e p r e d i c t i o n s i n t e r s e c t i n a s i n g l e p o s s i b i l i t y , i t i s h i g h l y probable t h a t t h e word ( o r s y n t a c t i c c l a s s ) s o predicted i s c o r r e c t . I f t h e p r e d i c t i o n s do not i n t e r s d c t , p a r s i n g is continued from t h e a c t i v e c b n f i g u r a t i o n s which were not t r i a d e a r l i e r because of t h e i r s c o r e s and from t h e c o n f i g u r a t i o n s which could begin c o n s t i t u e n t s a t t h e r i g h t and of an i s l a n d . This continued p a r s i n g i s an attempt t o f i n d a path which r e s u l t s i n a common p r e d i c t i o n acg#ss t h e gap. If t h a t t o o f a i l s , then t h e c o n f i g u r a t i o n s which were saved because they could lead u p a chain of PUSHes o r POPS t o new c o n f i g u r a t i o n s a r e t r i a d . If no p o s s i b i l i t i e s are l e f t t o t r y and t h e r e i s s t i l l no p r e d i c t i o n t o f i l l t h e gap, t h i s information i s noted, b u t i t does not d e f i n i t e l y mean that t h e i s l a n d s a r e incompatible, s i n c e i n some c a s e s t h e gap could a c t u a l l y be f i l l e d by two words i n s t e a d of one. SPARSER h a s two k i n d of p r e d i c t i o n st h o s e b h i o h seem h i g h l y l i k e l y a n d t h o s e w h i c h seemaless l i k e l y . A h i g h l y l i k e l y p r e d i c t i o n , s u c h as o n e w h i c h i s made from b o t h s i d e 3 of a small g a p , i s o u t p u t i n t h e form of a prooosal, w h i c h i s a r e q u e s t t o t h e rest of t h e s y s t e m t o f i n d a word meeting t h e r e q u i r e m e n t s of t h e p r o p o s a l . A p r o p o s a l c o n t a i n s : 1 ) t h e item b e i n g p r o p o s e d , w h i c h i s e i t h e r a p a r t i c u l a r word o r l i s t o f words (from a WAD a r c ) , o r a s y n t a c t i c olass ( f r o m a CAT a r c ) , @r N I L , m e a n i n g a n y word (from a TSI a~c ) 2 ) t h o l o f t a n d / o r r i g h t b o u n d a r y p o i n t b ) o f t h e item 3 ) a t e s t w h i c h t h e item m u s t s a t i s f y ( t h e c o n t e x t f r e e t e s t from b h e a r c ) 4) t h e c o n t e x t of t h e p r o p o s a l , i . e . t h e word m a t c h ( e s ) on t h e l e f t apd/or r i g h t s i d e of t h e item b a i n g p r o p o s e d . ( T h i s i s t o h e l p t h e l e x i c a l r e t r i e v a l c o m p o n e n t t a k e i n t o a c c e u n t p h o n o l o g i c a l phenomena w h i c h may o c c u r a c r o s s word b o u n d a r i e s . ) A l l p r e d i c t i o n s w h e t h e r o r n o t t h e y a r e c o n f i d e n t e n o u g h t o become p r o p p s a l s are o u t p u t o a s m o n i t o y s . A m o n i t o r i s a n o t i f i c a t i o n t o t h e c o n t r o l c o m p o n e n t t h a t i f a w o r d m e e t i n g t h e r e q u i r e m e n t s of t h e m o n i t o r i s somehow f o u n d ( p e r h a p s by t h e a c t i o n of a p r o p o s a l ) , i t may be a d d e d t o t h e t h e o r y . T h u s a m o n i t o r a c t s l i k e a demon w h i c h s i t s a t a p a r t i c u l a r p o i n t i n t h e word l a t t i c e a n d watches f o r t h e a p p e a r a n c e of a word m a t c h w h i c h i t c a n u s e . A m o n i t o r c o n t a i n s : 1 ) the item b e i n g m o n i t o r e d f o r ( g e n e r a l l y a s y n t a c t i c categcry, b u t may be a w o r d o r a t e s t ) 2 ) t h e l e f t o r r i g h t b o u n d a r y p o s i t i o n of t h e item b a i n g monitored far 3 ) a. t e s t w h i c h t h e item m u s t s a t i s f y (same a s f o r p r o p o s a l s ) 4 ) t h e t h a o r y w h i c h g e n e r a t e d t h e s o n i t o r 5 ) t h e a r c i n t h e grammar w h i c h w i l l p r o c e s s t h e item i f f o u n d 6 ) t h e c o n f i g u r a t i o n from w h i c h t h e p r e d i c t i o n was made 7) a s c o r e , i n d i c a t i n g r o u g h l y how i m p o r t a n t t h e monitor i s , i . e . how much i n f o r m a t i o n i s l i k e l y t o be g a i n e d by p r o c e s s i n g an e v e n t f o r t h a t m o n i t o r . ( N o t i c e t h a t m o n i t o r s w h i c h a r e s e n t t o t h e c o n t r o l c o m p o n e n t a r e v e r y much l i k e m o n i t o r s w h i c h a r e s e t i n t h e WFST by t h e o c c u r r e n c e of PUSH a r c s . ) Once t h e p r o p o s a l s h a v e b e e n m a d e a n d t h e m o n i t o r s h a v e b e e n set., SPARSER b u n d l e s u p t h e i n f o r m a t i o n i t knows a b o u t t h e c u r r e n t t h e o r y , s u c h as t h e c o n f i g u r a t i o n s a n d t r a n s i t i o n s i n the t h e o r y , a n y c o n f i g u r a t i o n s w h i c h a s t i l l c a n d i d a t e s for e x p a n s i o n , t h e c o n s t i t u e n t s i n t p o s a l s , and m o n i t o r s w h i c h h a v e b e e n c r e a t e d , etc. a n d a s s o c i a t e s t h e b u n d l e w i t h t h e t h a o r y number. T h i s i n s u r e s t h a t SPARSER w i l l b e able t o p i c k u p w h e r e i t l e f t o f f i f i t i s l a t e r g i v e n t h e t h m r y t o p r o c e s s f u r t h e r . P r o c e s s i n q M u l t i~l e T h e o r i e s T h u s f a r w e h a v e s e e n o n l y t h e o p S r a t i o n s w h i c h SPARSER p e r f o r m s on a s i n g l e t h e o r y , b u t w e made t h e a s s u m p t i o n t h a t SPARSER would b e g i v e n a number o f t h e o r i e s t o p r o c e s s i n s e q u e n c e . L e t u s now examine w h a t w i l l h a p p e n when the s e c o n d ( o r n t h ) t k e o~y i s p r o c e s s e d . Page 35 SPARSER w i l l no longer have a blank map and WFST; has been processed, result in^ i n the map shown i n Figure 4.6. Now we are going t o process a theory containin8 the island (4 NICKEL 16 22) (2 ANALYSES 22 30), which r e s u l t s i n the map of Figure 4 . 7 where the configurations and t r a n s i t i o n s added by t h i s theory are shown i n dotted l i n e s . #P/ADJ:16 and three possible lead-in t r a n b i t i o n s . Tho t r a n s i t i o n s for the two CAT N a r c s , however terminate on configurations which already existed i n the map, so the complete p a t h s from configuration NP/:16 t o configurations NP/N:30 and NP/N:22 w i l l b e discovered and processed, r e s u l t i n g i n the construction of two new noun phrases. Those new constituents would then r e s u l t i n t h e creation of configuration PP/PREP:16 and two new t r a n s i t i o n s . T h u s we have constructed only f i v e new configurations and seven new t r a n s i t i o n s and have been a b l e t o take advantage of s i x old configurations and s i x old t r a n s i t i o n s . I n t h i s fash-ion any information which has once been discovered about a possible parse path is made available t o anyother p a t h which can use i t .-No reparsinqi s ever done SPARSER merely r e a l i z e s the e x i s t e w e of relevant configurations and t r a n s i t i o n s and incorporates them i n t o the current theory.", "num": null, "type_str": "figure", "uris": null }, "FIGREF29": { "text": "Map a f t e r processing i s l a a d f o r \"nickel analysesif", "num": null, "type_str": "figure", "uris": null }, "FIGREF30": { "text": "i t i o n , a l l paths beginning w i t h t h a t t r a n s i t i o n which do not require t h e next word of i n p u t are a l s o included i n the s y n t a c t i c part of the theory.T h i s is accomplished by t r a c i n g from the terminating configuration a l l t r a n s i t i o n s w h i c h use e i t h e r tho same word of i n p u t a s t h e p r e v i o u s t r a n s i t i o n o r n o i n p u t word a t a l l .( A s i m i l a r p r o c e s s 19 u s e d t o t r a c e b a c k w a r d s , i.e. r i g h t t o l e f t , when n e e s s a r y . ) Uhen a c o n f i g u r a t i o n i s r e a c h e d w h i c h h a s n o t r a c e a b l e t r a n s i t i o n s e m a n a t i n g from i t , t h e t r a c i n g p r o c e s s , s t o p s . S i n c e b o t h transitions a n d c o n f i g u r a t i o n s a r e stored i n s u c h a way a s t o f a c i l i t a t e t r a c i n~ ( f o r example, each t r a n s i t i o n h a s a code a t t a c h e d t o i n d i c a t e w h e t h e r o r n o t i t c o n s u m e s o r tests i n p u t ) , t h i s p r o c e s s i s c o n s i d e r a b l y f a s t e r t h a n c r e a t i n g t h a t p o r t i o n of t h e map i n the first p l a c e . ( T o i l l u s t r a t e t h i s , a t h e o r y was p r o c e s s e d twice, o n c a w i t h a n e m p t y map a n d o n c a start in^ w i t h t h e map p r e v i o u s l y c r e a t e d ; t h e time r e q u i r e d f o r p r o c e s s i n g the t h e o r y f a l l from 4 7 . 5 s e c o n d s t d 16.5.) C o n f i g u r a t i o n s w h i c h c a n e n d trvaced p a t h s a r e p u t o n t h e a c t i v e c o n C i g u r a t i o n s list. I f , when o n e oP t h e m is p i c k e d u p f o r e x t e n s i o n , i t is d i s c o v e r e d t h a t t h e n e x t w o r d of i n p u t was used o n a t r a n s i t i o n a l r e a d y i n t h e map, t h e t r a c i n~ p r o c e s s i s r e p e a t e d . If t h e n e x t word of i n~u t i s new ( o r a t least h a s n o t c a u s e d any t r a n s i t i o n s from t h i ? c o n f i p u r a t i o n b e i n p c o n s i d e r e d ) then para in^ c o n t i n u e s i n ' t h e n o r m a l m a n n e r . P r o c e s s i h~ E v e n t s As-was m e n t i o n e d e a r l i e r , SPARSER can be c a l l e d upon t o a d d some new word m a t c h e s t o a t h e o r y i t h a s p r e v i o u s l y p r o c e s s e d . I n t h i s case, SPARSER i s s a i d t o process a n e v e n t . An a v a n t may be t h o u g h t of r a t h e r a b s t r a c t l y a s t h e d i s c o v e r y of a p i e c e of i n f o r m a t i o n t h a t h a s b e e n s y n t a c t i c a l l y p r o p o s e d , m o n i t o r e d f o r , o r n o t i c e d . C o n c r e t e l y , a n e v e n t is a p i e c e of d a t a c o n s i s t i n a o f : 1 ) t h e o l d t h e o r y t h a t p r o p o s e d o r s e t a monitor f a r t h e e v e n t 2 ) s o m e t h i n g t o be a d d e d t o the t h e o r y ( a new ward natch or c o n s t i t u e n t ) 4) t h e a r c i n t h e Rramnar w h i c h w i l l p r o c e s s t h e new i n f o r m a t i o n 4) t h e c o r i f i p u r a t i o n i h t h e o l d t b r o r y w h i c h will be a t one e n d of t h e t r a n s i t i o n c r e a t e d by t h e a b o v e a r c When SPARSER i s ~i v e n a n e v e n t , i t r e t r i e v e s f r o m its t a b l e s t h e b u n d l e o f c o n f i g u r a t i o n s , t r a n s i t i o n , e t c . i n t h e old t h e o r y . T h e n using the a r c a n d t h e n e w w o r d o r c o n s t i t u e n t i n t h e e v e n t , i t c r e a t e s t h e a p p r o p r i a t e t r a n s i t i o n c s ) . Then processing continues as u s u a l , that i s , a n y c o m p l e t e p a t h s are n o t i c e d a n d processed, a n d a n y new a c t i v e c o n f i~u r a t i o n s a r e e x b e n d e d , i f p o s s i b l e . New p r e d i c t i o n s may b e made as a r e s u l t of t h i s i n c r e a s e d i n f o r m a t i o n . ( A r e c o r d i s k e p t of p r e v i o u s p r e d i c t i o n s s o n o n e a r e remade u n l e s s w i t h a more l i b e r a l s c o r e . ) F i n a l l y SPARSER r e t u r n s t h e n G w , l a r g e r t h e o r y . T h i s new t h s o r y may be processed as p a r t o f a n o t h e r e v e n t a t some later time, t h u s g r a d u a l l y r e d u c i n g the number a n d size of t h e c a p s i n t h e t h e o r y . I f an e v e n t r e s u l t s i n f i l l i n g t h e f i n a l g a p i n a t h e o r y , and i f t h e r e s u l t i n p c o m p l e t e s e q u e n c e of w o r d s c a n be p a r s e d , SPARSER n o t i f i e s t h e c o n t r o l c o m p o n e n t o f t h i s f a c t , s i n c e t h e entire u t t e r a n c e may h a v e b e e n d i s c o v e r e d . O f c o u r s e , t h i s m a y n o t b e the correct s o l u t i o n -i t i s u p t o t h e c o n t r o l component t o l o o k a t m t h e a c o u s t i c ~o o d n e s s , s e m a n t i c m e a n i n g f u l n e s s , p r a g m a t i c l i k e l i h o o d , e t c . of t h e r e s u l t a s well a s t h e s y n t a c t i c s t r u c t u r e before d a c l a r l n g t h e u t t e r a n c e t o h a v e b e e n u n d e r s t o o d . I f f o r r e a s o n s o t h e r tham s y n t a c t i c , t h e u t t e r a n c e a p p e a r s t o be b a d , t h e c o n t r o l component of t h e system c o u l d ~o on t o t r y t o f i n d a n o t h a r , more s u i t a b l e , p o s s i b i l i t y . S e c t i o n 5 More D e t a i l s o f t h e Parsinfi Process 5.1 DEPTH vs BREADTH T h e parsing s t r a t e g y j u s t o u t l i n e d w o r k s b o t t o m u p when beg inn in^ t o p a r s e a n i s l a n d a n d when a c o n s t i t u e n t i s created w h i c h was n o t m o n i t o r e d f o r by t h e c u r r e n t t h e o r y . It works t o p dawn a f t e r an i s l a n d h a s b e e n s t a r t e d a n d t o make s y n t a c t i c predictions a t t h e e n d s of i s l a n d s . B o t h t o p down a n d bottom u p t e c h n i q u e s c a n b e e i t h e r d e p t h o r b r e a d t h f i r s t . Depth f i r s t p r o c e s s i n g takes a t e v e r y s t e p t h e ffl-st piece of i n f o r m a t i o n a v a i l a b l e a n d p u r s u e s i t s consequences. B r e a d t h f i r s t p r o c e s s i n g c o n s i d e r s a t e v e r y s t e p e v e r y p o s s i b l e n e x t s t e p o f e v e r y a l t e r n a t i v e and p u r s u e s a l l p a t h s i n p a r a l l e l . Breadth first p r o c e s s i n g g e n e r a l l y t a k e s much more space t h a n t h e d e p t h f i r s t many p a t h s would h a v e to be remembered a t o n c e Rags 40 i n s t e a d of h a v i n g j u s t o n e s t a c k w h i c h c o u l d b e popped a n d r e u s e d when n e c e s s a r y . T h e b r e a d t h f i r s t p r o c e s s m i~h t s a v e some computation steps a n d m i g h t produce several a m b i g u o u s p a r s i n~s s i m u l t a n e o u s l y w h i l e tba d e p t h first p r o c e s s would f i n d o n e before t h e o t h e r s ( t h e l a t t e r is a'small difference, s i n c e b o t h p r o c e s s e s w o u l d have t o be r u n t o e x h a u s t i o n t o i n s u r e t h a t a l l p o s s i b l e p a r s i n g s h a d b e e n f o u n d ) . I n p a r s i n g s p e e c h , some mixture of b r e a d t h f i r s t a n d d e p t h f i r s t p r o c~s s i n g can be e x t r e m e l y u s e f u l . To i l l u s t r a t e an a d v a n t a~e o f b r e a d t h f i r s t p r o c e s s i n g i n t h e s p e e c h e n v i r o n m e n t , c o n s i d e r w h a t m i g h t h a p p e n i f , d u r i n p t h e p r o c e s s i n 8 of a n i s l a n d t h e parser p i c k s u p a c o n f i r t i r a t i o n t o e x t e n d which h a s s e v e r a l p o s s i b l e arcs e m a n a t i n g from i t . If o n e a r c i s c h o s e n a n d a l l the o t h e r s a r e h e l d as a l t e r n a t i v s s ( i m e m d e p t h f i r s t ) , b u t t h e c h o s e n arc is w r o n p , a l l s u b s e q u e n t p a t h s beg inn in^ w i t h t h a t a r c w o u l d have t o b l o c k before the a l t e r n a t i v e s w o u l d be t r i e d . However, i f t h e e n d of t h e i s l a n d were r e a c h e d b e f o r e . t h e s u c c e s s o r f a i l u r e of t h e f i r s t choice were confirmsd, t h e only way t h a t b a c k u p would e v e r t a k e p l a c e w o u l d be t o h a v e o n e o r more e v e n t s a d d w o r d s t o t h e t h s o r y s o t h a t t h e p a t h c o u l d b e e x t e n d e d u n t i l i t f a i l e d . S i n c e t h e pap wou.ld be l i k e l y t o be f i l l e d by ( i n c o r r e c t ) w o r d s p r e d i c t e d by t h e e r r o n e o u s p a t h , o r by n o w o r d s a t a l l i f t h e ( i n c o r r e c t ) p r e d i c t i o n s were n o t s a t i s f i e d , i t is n o t a t a l l c l e a r how the process would a v e r know t o b a c k u p . T h i s problem c a n n o t be e l i m i n a t b i n a t o r i a l l y u n a c c e p t a b l e s o l u t i o n ) b u t i t c a n be m o d i f i e d t o a p r a a t e x t a n t by a j u d i o i o u s c o m b i n a t i o n o f d e p t h a n d b r e a d t h f i r s t p r o c e s s i n g t o f i n d t h e best p a t h , n o t just t h e f i r s t one, t h r o u g h t h e i s l a n d . T h i s \"bast p a t h w is n o t ~u a r a n t e e d t o be the correct o n e , so i t is p o s s i b l e t o c o n t i n u e process in^ by e x t e n d i n g p a t h s w i t h were suspended e a r l i e r . SPARSER h a n d l e s t h e p r o b l e m by a s s i g n i n g a score &o e v e r y c o n f i g u r a t i o n w h i c h r e f l e c t s t h e l i k e l i h o o d o f t h e p a t h w h i c h t e r m i n a t e s o n t h a t c o n f i~u r a t i o n t o be c o r r e c t . The s c o r e c a n a l s o be t h o u g h t of a s a m e a s u r e of how good t h a t e o n f i g u r a t i o n l o o k s i n r e l a t i o n t o o t h e r s a s a c a n d i d a t e f o r e x t e n s i o n . One q u e s t i o n which was p r e v i o u s l y l e f t u n a n s w e r e d , how a subset of t h e a c t i v e c o n f i s u r a t i o n s is c h o s e n f o r e x t e n s i o n , c a n now b e a n s w e r e d : t h e s u b s e t of m a x i m a l l y s c o r i n g c o n f i g u r a t i o n s is c h o s e n a t each step u n t i l t h e m a x i m a l score of a c t i v t ! c o n f i g u r a t i o n s b e g i n s t o f a l l . ( T h e s c o r e on a c o n f i g u r a t i o n a n d the score of a p a t h t e r m i n a t i n r o n t h a t c o n f i g u r a t i o n a r e t h e same thing -w e will use w h i c h ever t e r m i n o l o g y seems most n a t u r a l a t t h e time.) T h e result of this p r o c e s s i s a s o r t o f m o d i f i e d b r e a d t h first approach, w h e r e a t o n e s t e p a l l t h e a l t e r n a t i v e s a r e t r i e d but a t t h e n e x t s t e p o n l y th& best o n e s a r e c h o s e n f o r f u r t h e r e x t e n s i o n . T h i s is similar t o t h e b e s t -f i r s t p a r s e r d e s c r i b a d by P a x t o n i n [I81 b u t i t c a n be a p p l i e d t o t h e s o r t of p a r t i a l p a t h s w h i c h SPARSER generates r a t h e r t h e n r e q u i r i n g t h e perfect information r e s u l t i n g from a s t r i c t l y left t o r i p h t approach.The auccesg of t h i s method is d i r e c t l y dapandent on t h e r a l a t i v z accuracy o f t h e s c o r e s w h i c h a r e assigned t o t h e p a t h s .", "num": null, "type_str": "figure", "uris": null }, "FIGREF31": { "text": "Sevdral a t t e m p t s have b~e n made t o d*velop r i g o r o u s systems f o r p a r s i n g a r r o r f u l o r spedch-like i n p u t baaed on p r o b a b i l i t i e s [ I , 1 4 , 271. These attempts h a v e a l l s i m p l i f i e d t h e p r o b l e m t o such an axtent t h a t i t is no l o n p e r r e a l i s t i c o r e x t e n d i b l e , e.p. by assuming t h e i n p u t is a sequence ( r a t h e r t h a n a l a t t i c e ) of p r o b a b i l i t y d i s t r i b u t i o n s , by assuming t h a t a l l t h e n e c l s b a r y information is p r e s e n t i n t h e searhh space t o b e g i n w i t h s o t h e only problem i s t o find an o p t i m a l path throuph t h s s p a c z , by r e q u i r i n e a small v o c a b u l a~y , and/or by l i m i t i n g t h e Rranmar t o b e c o n t e x t free. T h e i d e a l s c o r i n g mechanism f o r SPARSER w o u l d be one w h i c h ac c u r a t e l y r e f l e c t e d a t every s t e p t h e p r o b a b i l i t y t h a t t h e p a t h8 c o r r e c t . Bayas r u l e could be Used, but i t would ba necessary t o know, a t any point i n t h s p a r s i n g process, what t h e p r o b a b i l i t y i s t h a t t h~ next a r c under c o n s i d e r a t i o n i s c o r r e c t , given that t h e entire p a t h u p t o t h e c u r r e n t s t e p i s c o r r e c t . I n o r d e r t o use t h i s a p p l i c a t i o n of Bayas r u l e i t would be necessary to p r a -c a l c u l a t e t h e p r o b a b i l i t i s s for evary p o s s i b l e p a t h and p a r t i a l p a t h which could b e generated -a c l e a r l y impossible t a s k s i n c s t h e r e are an i n f t n i t e number of such p a t h s .", "num": null, "type_str": "figure", "uris": null }, "FIGREF32": { "text": "Givzn t h a t w e c a n n o t c a l c u l a t e t h e p r o b a b i l i t i e s w e n e e d exactly, what is t h e next best o p t i o n ? If w e i g n o r e t h e e f f e c t of tila p a t h traversed u p t o t h e c u r r e n t p o i n t , b u t can say f o r r a y ~i v e n s t a t e how l i k e l y e a c h a r c em an at in^ from t h a t s t a t e is t o ba correct, we would h a v e a model which uses o n l y l o c a l Paforaation r a t h e r t h a n one w h i c h t a k e s i n t o a c c o u n t a c c u r a t e l yall1 tbe Left c o n t e x t w h i c h is a v a i l a b l e . Since it was n o t p r a c t i c a l t o r u n large amounts of d a t a t b m~h a parser i n o r d e r t o o b t a i d a c c u r a t e m e a s u r e m e n t s e v e n Sor t h e l i m i t e d model, t h e a u t h o r r e ' l i e d on c o n s i d e r a b l e experfeace w i t h ATN grammars t o a s s i g n a w e i g h t t o each arc o f t b a grqmmar r e p r e s e n t i n g t h o i n t u t i v e l i k e l i h o o d t h a t t h e arc f i i it can be t a k e n ) i s t h e correct cane t o c h o o s e from that state. These weights are small i n t e g e r s (0 t h r o u~h 5) -t h e jarger t h e w e i g h t the more l i k e l y t h e a r c . The question might a r i s e a s t o why t h e score o G t h e word sate4 used by a n a r c s h o u l d n o t b e u s e d t o i n f l u e n c e t h e score of tbu path using i t . SPARSER tries t o t r e a t e a c h t h e o r y as ladrependently as p o s s i b l e a n d tq a s s i g n scores based o n l y o n t h e syntactk i n f o r m a t i o n w h i c h is available. The o n e e x c e p t i o n t o t h i s rule is t h e s e m a n t i c i n f o r m a t i o n w h i c h is u s e d t o score constituents. If l e x i c a l Qord m a t c h scores were u s e d , t h e cbmtrol component would n o t b e a b l e t o separate t h e lexical goudwss from the syntactic g o o d n e s s of the t h e o r y a n d make Judgments aa t o t h e i r r e l a t i v e i m p o r t a n c z . I n a s y n t a x -d r i v e n apeecb u n d e r s t a n d i n g s y s t e m , h o w e v e r , i t would p r o b a b l y bz u s a f u l to combine l e x i c a l scores w i t h s y n t a c t i c i n f o r m a t i o n . As was d e s c r i b e d i n t h e p r e v i o u s s e c t i o n , when SPARSER b e g i n s t o p a r s e a n i s l a n d e a c h p o s s i b l e p a r t i a l p a t h is b e g u n by c r e a t i n g a c o n f i g u r a t i o n a t t h e h e a d of a t r a n s i t i o n f o r an a r c w h i c h can use t h @ c u r r e n t w o r d . R a t h e r a r b i t r a r i l y , i t was d e c i d e d t o g i v a t h i s c o n f i~u r a t i o q a score of o n e . T h i s starts a l l p a r t i a l p a t h s out e q u a l l y , a t e c h n i q u e w h i c h is n o t q u i t e accuratb, s i n c e some c o n t e x t s a r e more likely t h a n o t h m s . For e x a m p l e , t h e w o r d s t t t o m a n d \" f o r w a r e more l i k e l y t o o c c u r I np r e p o s i t i o n a l phrases t h a n i n s e n t a n t i a l complements.I f t h i s s i m p l i f i c a t i o n a p p a a r v t o h a r m t h e o ve r a l l p e r f o r m a n c e o f SPARSER, i t c o u l a be remadikcf by g i v i n g e a a h s t a t e an a p r i o r i s c o r e similar t o the weights o n a r c s . C o n f i g u r a t i o n s o n l e a d -i d p a t h s are a l s o g i v e n a score of o n e . After t h e i n i t i a l s t e p , w h e n a v e r a t r a n s i t i o n ( o t h a r than a PUSH or POP) is made, the score of the s u b s e q u e n t c o n f i g u r a t i o n is i n f l u e n c e d by t h e s c o r e of t h e c o n f ' i~u r a t i o n b e i n g e x t e n d e d and t h e w e i g h t o n t h e arc b e i n s u s $ d . If t h e s c o r e s were a c t u a l p r o b a b i l i t i e s ; t h e y would b e m u l t i p l i e d ; s i n c e t h e y a r e n o t , i t was a r b i t r a r i l y d e c i d e d t o a d d t h e m . When a t t e m p t i n g t o c r e a t e a c o n f i g u r a t i o n w h i c h already exists (a s i t u a t i o n e n c o u n t e r e d w h e n e v e r two o r more parse p a t h s f o r t h e same t h e o r y merge), t h s c o n f i g u r a t i o n is given t h e maximum of t h e s x i s t i n g s c o r e a n d t h e score w h i c h w o u l d h a w been assigned h a d t h e c o n f i g u r a t i o n been created anew. Whsn a PUSH a r c i s e n c o u n t e r e d a n d a c o n f i g u r a t i o n created t o b e g i n the s e a r c h f o r t h e r e q u i r e d c o n s t i t u z n t , t h e s c o r e of t h a t c o n f i g u r a t i o n is set t o b e t h e sum o f t h e s c o r a of t h e c o n f i g u r a t i o n c a u s i n g t h e PUSH, a n d t h e v a l u e (If a n y ) o f t h e l o o k -a h e a d t e s t on t h e PUSH a r c . F o r example, upon e n c o u n t a r i n r an a r c s u c h a s (PUSH NP/ ((NPSTART) T T ) ... ) t h e l o o k -a h e a d f u n c t i o n NPSTART r e t u r n s a h i g h integer valua ir t h e next word is a noun a n d a lowen v a l u a i f i t is a v e r b ( e . g . l t a c c o u n t i n g c o s t s t t ) . Of c o u r s e , i f tha l o o k -a h e a d f u n c r i o n fails a l t o g e t h e r , t h e c~n f i p u r a t i o n is n o t s e t u p , a l t h o u~h hhe m o n i t o r i n t h e WFST r e m a i n s . When a c o n s t i t u e n t i s c o m p l e t e d ( o r f o u n d i n t h e WFST) e n d a PUSH t r a n a i t i q n i s a b o u t t o b e made, t h e s c o r e o f the c o n f i~u r a t i o n on G h i c h t h e t r a n s i t i o n t e r m i n a t e s i s a f u n c t i o n ~f t h e score of t h e c o n f i~u r a t i o n heinc e x t e n d e d t h e weipht on t h e a r c , an@ t h e s c o r e of t h e c o n s t i t u e n t i t s e l f . T h e s c o r e of t h t * c o n s t i t u e n t i s c u r r e n t l y v e r y a d h o c , b d n g a f u n c t i o n of t h e number of w o r d s i n t h * c o n s t i t u e n t ( l a s s a f u n c t i o n o f t h e number of s u b -c o n s t i t u e n t s subsumed by this c o n s t i t u e n t , b o o s t e d i f t h e c o n s t i t u e n t is a mador o n e ) a n d the s c o r e w h i c h is d e t e r m i n e d by s e m a n t i c v e r i f i c a t i o n . T h u s s e m a n t i c a l l y ''roodIf c o n s t i t u e n t s w i l l s t the s c o r e s of t h e p a t h s which u s e t h e m more t h a n s e m a n t i c a l l y \" b a d w o n e s . Due t o t h e l e v e l of e f f o r t r e q u i r e d t o g a t h e r a c c u r a t e s t a t i s t i c s on t h e r e l a t i v e f r e q u e n c i e s o f a r c s , t h e c u r r e n t scores a r e a d m i t t e d l y a d h o c . I t is n o t c l e a r w h e t h e r d i f f e r e n t s c o r i n g m e c h a n i s m s would be b e t t e r , however i t is c l e a r t h a t t h e c u r r e n t s c o r i n g s t r a t e g y is b e t t e r t h a n n o s c o r i n g a t a l l , a s p r a i i m i n a r y m e a s u r e m e n t s i n d c a t e t h a t t h e number of t r a n s i t i o n s created ( a s well a s t h e number of c o n f i~u r a t i o n s a n d p r e d i c i o n s ) i s r e d u c e d a b o u t 25% b y t h z c u r r e n t s t r a t e g y . (It is r z a s o n a b l e t o a s k why semantic s c o r e s a r e u s e d t o i n f l u a n c e p a r s e p a t h s , s i n c d it was t u s t a r g u e d t h a t l e x i c a l Ycores s h o u l d not b e u s h d i n t h i s way S e m a n t i c scores may be more r e l i a b l e t h a n l e x i c a l o n e s b e c a u s e we a r e a s s u m i n g t h a t t h e u t t e r a n & is s e m a n t i c a l l y m a a n i b a f u l . Under t h i s a u s u m p t i o n , a c o n s t i t u d n t like \" r a n g e r e m a i n d e r \" as a noun-noun modifier a n a l o g o u s t o l t a u r p l u s moneyff s h o u l d b e r u l e d o u t a s e a r l y a s p o s s i b l e . S i n c e s u c h c o n 8 t i t u e n t s c a n n o t be r u l e d o u t on s y n t a c t i c rounds a l o n e , s i n c e p r o s o d i c i n f o r m a t i o n ( w h i c h m i g h t h e l p t o r u l e t h e m o u t ) is n o t a v a i l a b l e (see d i s c u s s i o n i n S e c t i o n 7.2), and s i n c e they w o u l d s e r i o u s l v o v e r r u n t h e p a r s e r with a p l e t h o r a of f a l s e p a t h s if t h e y wwe n o t r e J e c t e d , i t seems r e a s o n a b l e t o p e r m i t s e m a n t i c s t o i n f l u e n c e t h e p a r s e r . ) 5.3 SCORING P R E D I C T I O N S T h e p r e v i o u s s e c t i o n d i s c u s s e d t h r e e ways i n w h i c h SPARSER can make p r e d i c t i o n s a b o u t w h a t c o u l d f i l l i n g a p s b e t w e e n i s l a n d s . M o n i t o r s wait f o r the o c c u r r e n c e of a w o r d i n t h e word l a t t i c e ( o r a c o n s t i t u e n t i n t h e WFST), p r o p o s a l s r e q u e s t a s e a r c h f o r a p a r t i c u l a r s e t of w o r d s , a n d n o t i c e s i n d i c a t e t h e p r e s e n c e of a u s a b l e word i n t h e w o r d l a t t i c e ( o r a c o n s t i t u e n t i n t h e WFST). Since t h e p r o c e s s i n g of a t y p i c a l t h e o r y i s likely t o r e s u l t i n a number of p r e d i c t i o n s it is necessary t'o b e a b l e t o o r d e r t h e m s o t h a t p r e d i c t i o n s most l i k e l y t o b z c o r r e c t o r most l i k e l y t o y i e l d i m p o r t a n t i n f o r m a t i o n w i l l b e acted upon f i r s t . For example, i t is more important t o f i l l a Rap between two i s l a n d s t h a n t o extend a s i n~l e i s l a h d , s i n c e by f i l l i n g t h e Rap one can chzck t h e consistency of information which was l o c a l l y good i n aach i s l a n d i n d i v i d u a l l y b u t may not be c o n s i s t e n t when t h e y a r e j o i n e d . Since two words can occur t o~e t h a r i n (usually) many c o n t e x t s b u t l o n~e r szquances a r l generally more r e s t r i c t i v~, addine a word t o a one word i s l a n d is l i k e l y t o be l e a s p r o f i t a b l e i n terms of t h e number a,f p o s s i b l e p a t h s which are sliminatod by the a d d i t i o n than add in^ a word t o a multi-word i s l a n d . It is u p t o t h e s y n t a c t i c component t o i n d i c a t e t o t h e control component t h e r e l a t i v e importance a t t a c h s d t o each n o t i c a and monitor; the hipher t h e s c o r e , t h e s t r o n~e r t h e p r e d i c t i o n . Several f a c t o r s i n f l u e n c e t h e s c o r e attached t o p r e d i c t i o n s . One is t h e length of t h e i s l a n d t o w h i c h t h e p r e d i c t i o n is attached. One word i s l a n d s , i f they a r e processed a t a l l , yield v e r y l i t t l e information anc many p r a d i c t i o n s , bznca t h e p r e d i c t i o n s a r e not scored high. Proposals a r e l a s s important i f t h e r e is already a n o t i c s a b l e word i n the word l a t t i c e ( s i n c e t h a t word i s a c o u s t i c a l l y b e t t e r t h a n t h e word t o b e proposed, e l s e it would have bean found e a r l i e r . Howevsr, i f a proposal f i l l s a gap between two i s l a n d s , it i s given a h i g h e r s c o r e .Notices are boosted i n importance i f an e n t i r e c o n s t i t u e n t may be added a n d p e n a l i z e d i f t h e y w i l l add onto a one word i s l a n d .Scores range from 0 t o 95 f o r proposals, 0 t o 40 f o r n o t i c e s , and 0 t o 15 f o r monitors.", "num": null, "type_str": "figure", "uris": null }, "FIGREF33": { "text": "Theas s c o r e s appear t o work f a i r l y well with the rest of t h e BEN SPEECHLIS s y s t e m , but have b e e n dcvalopod by a process of i n t e r a c t i o n w i t h t h a o t h e r components ( i n o r d e r t o make t h e s c o r e s of s y n t a c t i c p r e d i c t i o n 8 commensurate w i t h t h o s e of semantic p r e d i c t i o n s ) and may b e chanazd considerably au t h e a t i r e s y~t e m e v o l v e s . Small s y n t a c t i c classes e m . detePml'naru and p r e p o s i t i o n s ) a r e proposed i n t h e i r e n t i r e t y ( t h a t i s , their elurnants a r e to be dnumrratad and ~i v d n t o the lexical matchinc componlnt f o r v e r i f i c a t i o n ) i f t h e i s l a n d w h i c h monitored for t h e m is more than one word l o n~. If a gap batwsen two i s l a n d 3 i s small e n o u c h f o r i u s t one word and i f a syntactic class h a s b e e n monitored f o r L from b o t h sides of the gap, i t i s propossd i r i t s e n t i r e t y a l s o . Sect ion 6 Examples and R e s u l t s SPAR3ER is w r i t t e n i n INTERLISP a n d r u n s on a PDP-10 u n d~r t h e TENEX o p e r a t i n g system . The program and i n i t i a l d a t a s t r u c t u r e s occupy approximately 90000 words of v i r t u a l memory. (The o t h e r componants of the BBN speech undzrstandicp system occupy s e p a r a t e f o r k s from t h e syntactic component.) Page 49 A t t h e time t h e ' e x a m p l e s i n t h i s s e c t i o n were r u n , the a l~o r i t h m c o n t r o l l i n g t h a d a c i s i o n -m a k i n g p r o c e s s i n t h e c o n t r o l c o m p o n e n t was u n d e r~o i n g r e v i a i o n a n d was n o t s o l i d i f i e d i n t o a f u n c t i o n w h i c h c o u l d operate a u t o m a t i c a l l y . R a t h a r , t h e r e ware a number o f p r i m i t l v d o p e r a t i o n s s u c h as scanning an u t t e r a n c e (or some s p e c i f i e d p a r t i o n of i t ) , c r e a t i n g t h o o r i e u , call in^ SPARSER w i t h a theory o r e v e n t , c a l l i n g f\"or t h e p r o c e s s i n g o f p r o p o s a l s , e t c . , w h i c h c o u l d b e i n v o k e d by a human s i m u l a t o r T h e follow in^ e x a m p l e s were p r o d u c e d i n t h i s mode, w i t h t h e u s e r act in^ a s t h e c o n t r o l c o m p o n e n t i n a way w h i c h c o u l d be m o d e l l e d by l a t e r i m p l e , m e n t a t i o n . S e v e r a l c o n v e n t i o n 8 have been u s e d i n t r a c i n~ t h e o p e r a t i f i g u r a t i o n w r i t t e n a s 30:NP/HEAD:23(39) is t h e c o n f i~u r a t i o n f o r s t a t e NP/HEAD a t p o s i t i o n 23 w h i c h h a s been g i v e n t h e ( u n i q u e ) n u m b e~ 30 and w h i c h c u r r e n t l y h a s a score of 3 9 . T h e", "num": null, "type_str": "figure", "uris": null }, "FIGREF34": { "text": "o t a t i o n s h a v c been inserted w i t h i n b r a c k e t s { 1; t y p e o u t i n u p p e r case was p r o d u c e d by t h e program. s e x a m p l e p a r a l l e l s t h a t g i v e n i n S e c t i o n F o u r . A word l a t t i c e was a r t i f i c i a l l y c r e a t e d w h i c h c o n t a i n e d o n l y t h e f o l l o w i n g t h r e e word m a t c h e s : t h i s v e r s i o n o r t h e s y s t e m , r a~u l a r i n f l e c t i o n a l e n d i n g s a r e i n c l u d e d i n word m a t c h e s after the element representing t h e s c o r e , h e n c e t h e somewhat p e c u l i a r word m a t c h f o r t h e w o r d \" t r i p s \" . ) Two t h e o r i e s were c o n s t r u c t e d , o n e f o r word m a t c h e s 2 a n d 3 , t h e o t h e r f o r 1 a n d 3. What f o l l o w s i s a n a n n o t a t e d (but otharwisc! u n e d i t , e d e x c e p t f o r c o n s i d e r a t i o n s o f s p a c i n g ) t r a n s c r i p t of SPARSER p r o c e s s i n g t h e s e two t h e o r i e s i n s e q u e n c e , u s i n g t h e MINIGRAMMAR of Figura 3.3 a n d A p p e n d i x I . SPARSER PROCESSING THEORY'I: 0 12 WINTER 16 TRIP -S 21 30 ( T h i s i s a l i n e a r r e p r e s e n t a t i o n o f the t h e p r y b a i n g p r o c e s s e d . T h e e n d p o i n t s a r e 0 a n d 30, b u t t h e w o r d s o c c u p y -o n l y t h e m i d d l e p a r t of t h e u t t e r a n c e . ) STARTING A N ISLAND \"WINTER\" TRYING CAT N A R C FROM NP/ADJ TO NP/ADJ { T h i s is t h e first of two arcs retrieved from ths i n d e f i r s t t r a n s i t i o n i s created, a n d since t h~r e is ,a CAT N a r c w h i c h e n t e r s s t a t e NP/ADJ, a m o n i t o r is s e t u p t o m o n i t o r f o r n o u n s w h i c h e n d a t p o s i t i o n 12(Now t h e l e a d -i n t r a n s i t i o n s a r e b e i n g c r e a t e d , a l o n g w i t h t h e m o n i t o r s f a r s y n t a c t i c c a t e g o r i e s w h i c h may p r e c e d e the n e w l y c o n s t r u c t e d c o n f i g u r a t i o n s . C o n f i g u r a t i o n s a l o n g t h e l e a d -i n p a t h a r e a l l a s s i g n e d a score of 1.3 MONITORING [ ADJ 1 JUMP TRANS # 3 FROM 4:NP/ART: 12( 1 ) TO 3:NP/QUANT: 12( ?) ENDING AT 12: M O N I T O R I N G [ ART 1 JUMP TRANS #4 FROM 5:NP/:12(1) TO 4:NP/ART:12(1) { T h e l a a d -i n t r a n s i t i o n s a r e a l l made. Now t h e s e c o n d a r c w h i c h can uae t h e noun i a a b o u t t o b e p r o c e s s e d . ) \"WINTER\" TRYING CAT N ARC FROM NP/ADJ TO NP/N CAT N TRANS #5 FROM 1:NP/APJ:12(1) TO 6:NP/N:16(6) { T h i s i s t h a s e c o n d of t h d two a r c s o b t a i n e d from t h e i n d e x t a b l e f o r \" w i n t e r w . Th* l e a d -l n t r a n s i t i o n s t o c o n f i g u r a t i o n 1 havd a l r e a d y b e e n C o n s t r u c t e d , s o t h r e a d y t o c h o o s e c o n f i g u r a t i o n s t o e x t e n d . T h e pool of c a n d i d a t e s f o r extension c o n t a i n s c o n f i~u r a t i o n s 2 a n d 6 . )", "num": null, "type_str": "figure", "uris": null }, "FIGREF35": { "text": ". a c t i o n is t a k e n a b o u t s t a r t i n g a c o n f i g u r a t i o n f o r s t a t e PP/ b e c a u s e t h e l o o k -a h e a d t e s t w h i c h c h e c k s t h a tt h e n e x t word c a n b e g i n a p r e p o s i t i o n a l p h r a s e fails on t h e word t r i p . ) TRYING POP ARC POP TRANS 86 FROM 6:NP/N:16(6) { C r e a t i n g t h s POP t r a n s i t i o n c o m p l e t e s a g a t h from c o n f i g u r a t i o n 5. Thz p a t h is e x p r e s s e d as a l i s t o f t r a n s i t i o n n u m b e r s . W e a r e a b o u t t o -e x e c u t e t h e p a t h , t h a t is, c h e c k t h e c o n t e x t -' s e n s i t l v s t e s t s a n dd b t h a r e~i s t e r build in^ a c t i o n s a l o n g i t . ) EXECUTING PATH (4 3 2 5 6) B E G I N N I N G AT TRANS 4, CONFIC 5 {We m u s t b a~l n a x e c u t i n p the p a t h a t t h e first t r a n s i t i o n , b e c a u s e n o p a r t o f i t h a s b e e n e x e c u t e d before. Later we w i l l see t h a t it is p o s s i b l e t o b e g i n e x e c u t i o n of a p a t h i n t h e m i d d l e , since t h * r e p i s t e r c o n t e n t s are s t o r e d a t e a c h s t e p . ] DOING J U M P ARC FROM 5:NP/ TO 4:NP/ART DOING JUMP ARC FWM 4:NP/ART TO 3:NP/QUANT DOING JUMP A R C FROM 3:NP/QUANT TO I:NP/ADJ DOING CAT ARC WITH WINTER FROM 1 :NP/ADJ TO 6: N P / N DOING POP ARC FROM 6:NP/N TEST FAILED {Tha t e s t failed b e c a u s e thzre is n o d e t e r m i n e r , a n d MINIGRAMMAR r e q u i r e s t h a t s i n g u l a r , u n d e t e r m i n e d n o u n s c a n be c o m p l e t e n o u n p h r a s e s o n l y i f t h e y are mass n o u n s . \"Winterv is not marked a s a mass noun i n o u r d i c t i o n a r y , hence i t w i l l n o t parse as a complete n o u n p h r a s e . ) SELECTED CONFIGS ( 2 ) FOR EXTENSION { S i n c e o x t a n d i n g c o n f i~u r a t i o n 6 d i d n o t do much for us, we g o b a c k t o t r y t h a lower s c o r i n g c o n f i~u r a t i o n 2 f i g u r a t i o n s , 7 a n d 8 , i s c h o s e n . ) UP C O N F I G ? : P P / r 2 1 ( 8 ) MONITORLNG [ PREP ] (Since there i s no n e x t word t o t e s t , a c o n f i p u r a t i o n is s e t u p t o b e g i n processing a p r e p o s i t i o n a l p h r a s e , and the? s y n t a c t i c c a t o g o r i e s ' w h i c h @ a n b e~i n s u c h a p h r a s e -i n t h i s c a s e , o n l y o n e -a r e m o n i t o r z d f o r . ) e a t i o n of t h e POP t r a n s c o m p l e t e d a p a t h , t h e f i r s t p a r t of w h i c h h a s a l r e a d y -b e a n e x i c u t i d . r e f o r e p i c k u p i n the m i d a d s . of' t h e p a t h a n d e x e c u t e o n l y t h e last t h r e e t r a n s i t i o n s . t h s u c c e e d s -n o d e t e r m i n e r i s n e e d e d s i n c e t h e h e a d noun i s p l u r a l -a n d a c o n s t i t u e n t is c o n s t r u c t e d . T h e s e m a n t i c c a m p o n e n t h a s b e e n t u r n e d e f f f o r t h i s e x a m p l e , s o it a d d s n o t h i n~ t o t h e acore r:hich SPAzSZS w s s i u n $ --5 p o i f i t s f o r each w o r d i n the c o n s t i t~l e n t .~ SYIJ WEIGHT + SEM WT = 10 + 0 = 10 {No m o n i t o r e x i s t s i n the WFST f o r a NP/ a t t h i s p l a c z , s o t h e a r c s ( i n N I N I G R A M M A R t h e r e i s o n l y one) w h i c h c o u l d p u s h f o r a NP are p r o c e s s e d bottom u p i n e x a c t l y t h e same m a n n e r as t h e two a r c s w h i c h c o u q d u s e a noun a t the b e g i n n i n p of t h e i s l a n d . ) NP/ WAS NEVER PUSHED FOR PUSH NP/ TRANS #I0 FROM 10:PP/PREP:12(", "num": null, "type_str": "figure", "uris": null }, "FIGREF36": { "text": "p o s a l s , monitors, and notlces constructed. The s y n t a c t i c w o r e assigned t o t h e theory i s ~i v o n -here just t h e acore of the conutituent constructed. Then there i s a summary of ~t a t i s t i c s . ) 12, CONFIG 1 DOING CAT ARC WITH SUMMER FROM 1:NP/APJ TO 2:NP/ADJ DOING CAT ARC WITH TRIP FROM 2:NP/ADJ TO 7:NP/N +*+u -DOING POP ARC FROM 7:NP/N MADE #2 FROM 12 TO 21 : SEM WT = 10 + 0 = 10 NP/ WAS PUSHED FOR AT C O N F I G 10 { T h i s time there a r e m o n i t o r y i n t h e WFST, o n e w h i c h is l o o k i n g f o r a NP start in^ a t p o s i t i o n 12 a n d o n e which i s l o o k i n g f o r a NP ending at position 21. One t r a n s i t i o n i s s u f f i c i e n t t o s a t i s f y b o t h of these, ant! t h e p r e p o s i t i o n n e e d e d t o c o m p l e t e a PP/ is m o n i t o r e d f o r . ) PUSH NP/ TRANS-6 1 3 FROM TO: P P / P R E P : 12( 1 ) TO tl:PP/NP:21(8) NPI MAY LEAD TO CONFIC 1 1 { T h i s i s caused by t h e fact t h a t there was a monitor T o r a , noun p h r a s e ending a t c o n f i p u r a t i o n 1 1 -t h e o n e craated when c o n s t i t u e n t 1 was m a d e . T h e t r a n s i t i o n w h i c h w o u l d bd 8 a t u p is t h e t r a n s i t i o n Just c r e a t e d , s o i t i s n o t remade. A l l of t h e p r o c e s s i n p w h i c h r e s u l t e d from the c o m p l e t i o n of a c o n s t i t u e n t i s f i n i s h e d ; h o w e v e r there a r e h o n i t o r s s t i l l t o be s e t f o r c o n f i g u r a t i o n s a l o n e the p a t h . ) n c e each m o n i t o r c o n s i s t 8 o f thz item b e i n c m o n i t o r e d f o r , its a s s o c i a t e d t e s t ( i f any) t h~ t h e o r y w h i c h i s t o be n o t i f i e d when t h e m o n i t o r is s a t i s f i e d , a n d the c o n f i g u r a t i o n a n d a r c c a u s i n g t h e m o n i t o r , m o n i t o r s m u s t b e mad* a n e w each time one of t h e e l e m e n t s changes, a l t h o u g h some of t h e list s t r u c t u r e c a n be shared, h e n c e t h z s e e m i n g p r o l i f e r a t i o n o f m o n i t o r s . ) {N80w SPARSER p r o c a s s s s the othsr a r c w h i c h c o u l d uss the word \"summerw. ) 14, C O N F I G 1 DOING CAT ARC WITH SUMMER FROM 1:NP/ADJ TO 6:NP/N DOING FOP ARC FROM 6:NP/N TES? F A I L E D i&cauae, \"sum a p n c a n n o t b e a complete noun p h r a s e i n i s t r a n s i t i o n was c r e a t e d b e f o r e , b u t i s now m a d s p a r t of t h e c u r r a n t theory. I t d o a s not c o m p l l t e a pqth or cause a n y f u r t h e r a c t i o n . I f i t h a d a t r r m i n a t i n f f c o n f i g u r a t i o n , i.e. i f a t r a n s i t i o n o t h e r t h a n a POP m i n a t i n p configuratdon would have been p l a c e d on t h e l i s t of p o s s i b l e c o n f i p u r a t i o n s t o e x t e n d . } SELECTED CONFIGS ( 6 ) FOR EXTENSION PICKING UP CONFIO 6:NP/N:e16(6) WITH WORD T R I P TRACING POP TRANS 6 FPOM 6 :~~/~: 1 6 ( 6 ) SELECTED CONFIGS ( 2 ) FOR EXTENSION PICKING UP CONFIG 2 : N p / A~J : 1 6 ( 3 ) WITH WORD TRIP TRACING CAT N TRANS 8 U S I N G \"TRIP\" FROM 2:NP/ADJ:16(3) TO 8 : N P / A D J : 2 1 ( 5 ) STARTING AT 2 1 : MONITORING [ N ] MONITORING [ N ] { T h a r z are two noun arcs L e a v i n g s t a t e NP/ADJ, h e n c e two m o n i t o r s . ) SELECTED CONFIGS ( 8 ) FOR EXTENSION PICKING UP CONFIG 8 : N P / A D J : 2 1 ( 5 ) ALL ARCS TRIED AT T H I S CONFIG PREDICTIONS:", "num": null, "type_str": "figure", "uris": null }, "FIGREF37": { "text": "on i n p u t w h i c h i s analogous t o t h e example p r e s e n t e d w i t hi l l u s t r a t i o n s of t h e map i n Section Four. T h e I n t e r e s t e d r e a d e r i s u r g e d t o draw his own maps w h i l e reading t h e follow in^ Pact! 56 examples i n o r d e r t o best u n d e r s t a n d t h e d y n a n i c o p e r a t i o n of This e x a m p l e is more r e a l i s t i c t ban t h e previous cntb -i t shows t h e o p e r a t i o n of SPARSER i n t h e c o n t e x t of an pt terance w h i c h h a s b e e n 2 u t o r n a t i c a l l y segmented a n d l a b e l e e , ~i t h the o n s t r a t e s how SPARSER c a n h e l p t o select the best s e t v f words from a o r c o m p l e x word lattict.", "num": null, "type_str": "figure", "uris": null }, "FIGREF38": { "text": "EXTENSION {This time two a c t i v e configurations have the same maximal score, s o they are both processed.) P I a f f f i UP COIMG 3:NP/:0(11) WITH WORD WHAT TPTIYC GAT QDET ARC a CAT QOET TRANS #3 FROM 3:NP/:0(11) TO 5:NP/ORD:3(16) ma= UP CONFIG ~: s / N P :~( I I ) STAPIIMG A t 3: H M I T O~l r G [ MODAL 1 N G JUMP A R C WITH WHAT FRdlY l:S/ TO 2:S/Q DOING CAT ARC WITH WHAT FROM 21S/Q TO 4:SINP D O I N G POP ARC FROM Q:S/WP TEST FAILED { T h i s t e s t f a i l e d b e c a u s e the crammar does n o t a l l o w 'lwhatw t o b e a complete s a n t e . n c a . ) SELECTED CONFIGS (5) F O R EXTENSION P I C K I N G UP CONFIG 5:NP/ORU:3(16) STARTING AT 3:MONITORING [ QUANT/ ]SETTING UP CONFIG 6: QUANT/ : 3 ( 1 6 ) (Hare all the words which c a n s t a r t q u a n t i f i e r s , like \"a h un d r e d w o r \" p o i n t f i v a m , ard p r o p o s e d . T h e grammar d o e s n o t p r e c l u d e a q u a n t i f i e r f o l l o w i n g a q u e $ t i o n -d e t e r m i h z r , e -g . \"What t h r e e men traveled t o Spain'llI. ) {MONITORING [ INTEGER ZERO NO POINT A ] For c o n s i d e r a t i o n s of s p a c e , l o n g listings of m o n i t o r s a n d p r o p o s a l s i n t h i s e x a m p l e will b e c o m p a c t e d as shown here. Such a l t e r a t i o n s t o the a c t u a l trace produced w i l l be s u r r o u n d e d b y b r a c k e t s . ) TRYING JUMP NP/QUANT ARC JUMP T R A N S #5 FROM 5:NP/ORD:3(16) TO 7:NP/QUANT:3(21) SELECTED CONFIGS ( 7 ) FOR EXTENSION PICKING UP CONFIG 7 : NP/QUANT :,3 ( 2 1 ) TRYING JUMP NP/DET ARC JUMP TRANS 86 FROM 7:NP/QUANT:3(21) TO 8 : N P I D E T : 3 ( 2 6 ) SELECTED CONFIGS ( 8 ) FOR EXTENSION PICKING UP CONFIG 8 : N P / D E T : 3 ( 2 6 ) STARTING AT 3: MONITORING [ NPR/ NPP/ 1 { T h e r e a r e two PUSH N P R / a r c s from t h i s s t a t e s o two m o n i t o r s a r e c r a a t e d , b u t only one c o n f i g u r a t i o n i s s e t u p -1 SETTING UP CONFIG g:NPR/:3(26) {MONITORING [NPR NPR N A D J N V ADV]) TRYING JUMP NP/HEAD A R C JUMP TRANS #7 FROM 8:NP/DET:3(26) TO 10:NP/HEAD:3(29) SELECTED CONFIGS ( 1 0 ) FOR EXTENSION PICKING UP CONFIG 10:NP/HEAD:3(29) {This i s a n e x~m p l e of he f a l l i b i l i t y of using o n l y c o n t e x t free tests on p a r t i a l p a t h s , T h e parser thinks i t has s u c c a s s f u l l y reached state NP/HEAD, while i n f a c t t b i s c a n n o t b e t h e case b e c a u s e n o h e a d nouh h a s b e a n d l s c o v a r a d f o r the noun p h r a s e . Thus it i s i n c o r r e c t t o WITH WHQ FROM 3:NP/ TO 5:NP/ORD DOING JUMP ARC FROM 5:NP/ORD TO 7:NP/QUANT DOING JUMP ARC FROM 7:NP/QUANT TO 8:NP/DET DOING JUMP ARC FROM 8:NP/DET TO 10:NP/CAT N A R C FROM NP/DET TO NP/HEAD CAT N TRANS # J 5 FROM 14:NP/DET:6(1) TO 21:NP/HEAD:19(6) {This arc i s u s i n g ltregistrationtl as t h e head noun o f a noun phrase. ) SELECTED CONFIGS ( 2 1 ) FOR EXTENSION PICKING UP CONFIG 2 1 : NP/HEAD: 19( 6 ) STAHTING AT 10: FIONIT nxwc [ R / PP/ W / N I L I SETT 9 NG UP CONFIG 2 2 : R / : l Q ( 6 ) {MONITORING [PREP WHOSE \\,!I10 W H I C H THAT WHOM] ) t i c e i s i n r e s p o n s e t o t h e look-ahead test o n t h e p u s h a r c t o a t a t e R/NIL. S i n c e 'vfee'l c a n s t a r t a r e d u c e d r e & a t i v e c l a u s e , i t i s n o t i c e d , b u t t h e r e is n o t a specific m o n i t o r s e t u p b e c a u u e t h e a r c w i t h i n t h e r e l a t i v d c l a u s e network w h i c h will a c t u a l l y p r o c e s s t h e w o r d I t f e d \" i s n o t k n o w n . ) TRYING POP ARC POP TRANS # 1 6 FROM 21:NP/HEAD:19(6) EXECUTING PATH ( 1 4 1 3 12 1 1 1 0 1 5 1 6 ) BEGINNING AT TRANS 1 4 , CONFIG 20 TEST FAILED { T h e p a t h f a i l e d b e c a u s e t h a r e is n o d e t e r m i n e r f o r t l r e g i s t r a t i o n . \" 1 PREDICTIONS:", "num": null, "type_str": "figure", "uris": null }, "FIGREF39": { "text": "N A D J ) FROM 3 TO 6 { P r o p o s a l s were made t o f i l l t h e g a p because t h e r e were m o n i t o r s from b o t h sides of a pap small e n o u s h t o c o n t a i n o n e w o r d . ) FINISHED THEORY 1 WITH S Y N SCORE 0 { I t took 1 1 . 9 s e c o n d s t o p r o c e s s t h i s t h e o r y . ) { P r o c e s s i n g t h e proposal-s j u s t made r e s u l t s , n o t a b l y , i n t h e d e t e c t i o n of t h e w o r d l v o t h e r v b e t w e e n v v~h a t v l a n d w r e~i s t r a t i o n l v , b u t t h e word match score i s v e r y l o w . Word matches f o r *isw a n d Itare\" from p o s i t i o n 3 (next t o \" w h a t v v ) t o p o s i t i o n 4 are a l s o f o u n d , b u t s i n c e t h e y do not fill t h e pap, t h e event scores a r e low. The b z s t event is that for the word \"faen. P r o c a s s i n~ it i s fairly uninteresting, since it completes no constituent, so we w i l l omit the trace of that event. After it hasbean processed, however, t h e best event is that for the word \" t h e v f and the theory just created. ] S Y N T A X P R O C E S S I N G E V E N T F O R T H E O R Y # 2 WITH NEW WORD (4 THE 6 ) T O GET NEW T H E O R Y # 3 . 0 WHAT 3 4 T H E 6 R , E G I S T R A T I O N 19 FEE 23 \"THE\" T R Y I N G ( C A T A R T --) F R O M S T A T E N P / O N L Y T O C O N F I G 18 CAT ART T R A N S #26 F R O M ~~: N P / O N L Y :~(~)TO 18:NP/ART:6(6) ENDING A T 4:M O N I T O R I N G [ O N L Y ]P R O P O S I N G \"ONLY\" J U M P T R A N S #27 FROM 3 3 : N P / : 4( 1 )T O 32:NP/ONLY:4(3) E X E C U T I N G PATH", "num": null, "type_str": "figure", "uris": null }, "FIGREF40": { "text": "NU. S G **++ ha format o f this noun p5rase is slightly different from that in the previous example because tpe ~t r u c t u r e building action for noun phrases in S P E E C H G R A M M A R is different from that in M I N I G R A M M A R .", "num": null, "type_str": "figure", "uris": null }, "FIGREF41": { "text": "n u m b d r of predictions and notices. )SYN WEIGHT+ SEM WT = 15 + 0 = 15 NP/ W A S N E V E R PUSHED FOR PUSH NP/ T R A N S #28 F R O M 3 4 : F O R / F O R : 4 I T O R I N G [ PREP ] NP/ WAS N E V E R P U S H E D F O R P U S H N P / T R A N S 130 F R O M 3 8 : R / N I L : 4 ( 1 )T O 3 9 : S / N P : 2 3 ( 9 )N P / W A S N E V E R P U S H E D F O R", "num": null, "type_str": "figure", "uris": null }, "FIGREF42": { "text": "TRYING (CAT V --) FROM CONFIG 4 CAT V TRANS #60 FROM 4:S/NP:3(11) TO 45:S/AUX:4(16) {This transition does not immediately complete any paths, s o t h e best scoring configurations of the theory are EXTENSION PICKING U P CONFIG 59:S/VP:23(12) TRYING JUMP S/S ARC JUMP TRANS C66 FROM 59:S/VP:23(12) TO 61:S/S:23t h i s thzory iq very s i m i l a r to t h a t of the f i r s t t h e o r y in t h e p r e v i o u s ~x a m p l e , and w i l l not be commented upon here. The p u r p o a e in show in^ i t is t o provide a map, part of which t h e n e x t c a l l t o SPARSER will t r a c e . ) STARTING AN ISLAND STARTING AT LEFT END OF SENTENCE SELECTED CONFIGS ( 1 ) FOR EXTENSION PICKING UP CONFIG 1 : S / : 0 ( 1 ) WITH WOHD WHAT T R Y I N G PUSH P P / A R C TRYING JUMP S/Q A R C JUMP TRANS # I FHOM 1 : S / : 0 ( 1 ) TO 2:S/Q:0f6) T R Y I N G WRD I F A H C TRYING JUMP S/IblP ARC TRYING JUMP S / D C L ARC SELECTED CONFIGS ( 2 ) FOR EXTENSION PICKING UP CONFIG 2 : S / Q : 0 3 : N P / : 0 ( 1 1 ) WITH WORD WHAT TRYING WRD ONLY ARC TRYING CAT QDET ARC CAT QDET TRANS # 3 FHOM 7 : N P / : O ( l l ) TO 5 : N P i O R D : 3 ( 1 6 ) TRYING PUSH DATE/ A N G JUMP ARC WITH WHAT FROM l : S / TO 2:S/Q DOING CAT ARC WITH WHAT FROM 2:S/Q TO 4:S/NP D O I N G POP ARC FROM 4:S/NP TEST FAILED SELECTED CONFIGS (5) FOR EXTENSION PICKING UP CONFIG 5 : N P / O R D : 3 ( 1 the w o r d lattice, hence two n o t i c e s are c r e a t e d . ) MONITORING [ V 1 JVMP TRANS #I0 FROM 16:NP/QUANT:19(1) TO 14:NP/DET:19(1) E N D I N G AT 19: t N TOR NG W A N T / ] J U~ IRAN^ 1 FROM 17:NP/ORD: 19( 1 ) TO 16:RP/QUANT: 19( O N F I G 54t S/VP: 19 ( 12) TRYING J U M P $ I S AHC JU!lP TRAMS #49 FROll 5 4 : S / V P : l Q ( 1 2 ) TO 5 5 : S / S : l a ( l l l ) Sl2LECTED CONFIGS ( 55 ) FOR KXTlPaths~E o r i n p Predictions6 Examples a n d ResultsExample 1Examole 2Example 37 Conclus~onsS t r e n g t h s PrasodicsExtensions and F u r t h e r ResearchConclusionA p p e n d i x I", "text": "", "html": null, "type_str": "table", "num": null }, "TABREF2": { "content": "
0 26111422303538
GIVE EIGHTY PEOPLE DONETENMODAL DETERMINATION ROCK
0361114 1518 21263538
WERE ANY PEOPLEMETALSEVEN
0361117212732
F i j i u r e 4 . 1
", "text": "", "html": null, "type_str": "table", "num": null }, "TABREF4": { "content": "
4(1) PUSH NP/ TRANS #43 FROM 51:VP/NP:J+(1) TO ~~: V P / V P : Z~( ? ) ENDING AT 4 : !lONITORING THERE J PROPOSING \"THEREvf NP/ WAS NEVER PUSHED FOR ENDING AT 4: MONITORING [ NP/ 1 JUMP TRANS #44 FROM 47:VP/HEAD:4(1) TO ~I:VP/NP:~(I) BEGINNING A? TRANS 15, CONFIG 14 DOING CAT ARC WITH REGISTRATION FROM 14:NP/DET TO 2l:NWHEAD DOING POP ARC FROM 21tNP/HEAD **+* MADE #2 FROM 4 TO 1 3 : NP DET ART THE N REGISTRATION FEATS NU SG * # * * {This constituent can now satisfy the monitors set by the discovery of the largar one, result in^ in the creation of many new transitions but no new predictions.) S Y N PUSH NP/ TRANS C46 FROM 34:FOR/FOR:4(7) TO 53:T0/:19(8) {Similar tIP/ transitions are ast u p at configurations 36,38,40,41,43,46,47,49, PICKING UP CONFIG 55:S/NP:19(8) WITH WORD FEE TRYING POP ARC POP TRANS 856 FROM 55:S/NP:19(8) E X E C U T I N G PATH (48 5 6 ) BEGINNING AT TRANS 40, CONFIG 38 TEST FAILED E X E C U T I N G PATH ( 3 3 50 56) BEGINNING AT TRANS 33, CONFIG 42 ++** MADE # 3 FROM 4 TO 19: S NPU NP DET ART THE ADJ NP N REGISTRATION NU SG N FEE FEATS NU. SG WITH FEATURES (NPU) **** NP/ EXECUTING PATH (27 26 12 1 1 10 15 16) {Here
", "text": "WAS NEVER PUSHED FOR PUSH NP/ TRANS #45 FROM 49:V~/V:4(1) TO 44:VP/V:23(9) {The WEIGHT + SEM WT = 10 + 0 = 10 NP/ WAS PUSHED FOR AT CONFIG 34", "html": null, "type_str": "table", "num": null }, "TABREF5": { "content": "
)
PICKING UP CONFLG 54:PP/NP:19(8) WITH WORD FEE
TRYING POP ARC
POP TRANS #57 FROM 54:PP/N~:1?(8)
P I C K I N G UP CONFIG 39:S/NP:23(9)
TRYING POP ARC
POP TRANS #58 FROM ~~: s / N P : ?~( ! J )
EXECUTING PATH (30 58)
BEGINNING AT TRANS 30, C O N F I G 38
TEST F A I L E D EXECUTING PATH (33 32 58)
****REGINNING AT TRANS 3 2 , CONFIG 41
MADE #4 FROM 4 TO 23:
S NPU
NP ADJ NP N REGISTRATION
NU SG
DET ART THE
N FEE
FEATS NU SG
****WITH FEATURES ( N P U )
PICKING UP CONFIG 37:PP/NP:23(10)
TRYING POP ARC
POP TRANS 659 FROM 3 7 :~~/~P : 2 3 ( 1 0 )
PREDICTIONS:
NOTICING ( 1 9 IS 3 4 -79 0), SCORE 10
NOTICING (20 ARE 3 4 -128 O), SCORE 10
PROPOSING (ONLY FOR WHICH
", "text": "SYN WEIGHT+ SEM WT = 15 + 0 = 15 {MONITORING [ ONLY FOR W H I C H THAT WHO WHOM THERE 1 ENDING AT 4, SCORE 15 MONITORING [ MODAL NEC V QADV PARTICLE", "html": null, "type_str": "table", "num": null } } } }