{ "paper_id": "M95-1008", "header": { "generated_with": "S2ORC 1.0.0", "date_generated": "2023-01-19T03:12:44.177293Z" }, "title": "Knight-Ridder Information's Value Adding Name Finder : A Variation on the Theme of FASTU S", "authors": [ { "first": "Arkady", "middle": [], "last": "Borkovsky '", "suffix": "", "affiliation": {}, "email": "arkady@dnt.dialog.com" } ], "year": "", "venue": null, "identifiers": {}, "abstract": "Knight-Ridder Information, Inc., participated in MUC-6 with VANF (Valu e Adding Name Finder), the system used by Knight-Ridder Information i n production for adding a Company Names descriptor field to online newspape r and newswire databases. Knight-Ridder Information participated in the NE tas k only. The system used for MUC-6 is exactly the same as is used in production. The only difference is the input and output formats. VANF used a cascaded non-deterministic state machines approach and is based on FASTUS. 1. Backgroun d Late 1992, we realized that a company like Knight-Ridder Information (Dialog Information Services, a t that time) had to get into the business of Value Added Reselling of information. An obvious question was: what kind of value can be added to the information? The obvious answer was that the added value shoul d have the same nature as the original : it should be information. The plan was to introduce NLP into Dialog's technology to add informational value to raw documents. The industrial environment required that anything we work on has a fmite development time and is usable in production after the developmen t is over. This dictated step-wise development and required starting with solvable problems. The obvious choice was smart tokenization, which included named entity detection. Originally, we tried to buy the technology from outside, so that a full-fledged partnership could b e developed based on the initial project. Two first attempts ended in nothing (although they were quite useful experiences for Dialog, and probably for the potential partners as well). The third one, with SRI , led to development of the current system, VANF (Value Adding Name Finder), which is routinely used i n production at Knight-Ridder Information. VANF is an independent re-implementation of FASTUS [I]. At first, we wanted to use FASTUS as it was. However, the production requirements prohibited using a Lisp environment, and we decided to port it (to c++). A direct port was not feasible ; at the same time, SRI decided that a declarative, grammar-like description of FSM was more intuitive and easier to work wit h than the graphical tools they were using at the time. So a grammar definition language was defined, w e spent some time discussing and refining it, and I implemented an engine for Dialog's version of the language. Under the contract we had, SRI delivered a transcription of FASTUS rules in the new (declarative) language, although they did not have their own interpreter (nor even a complete languag e definition) at that time. Somehow, we made this transcription compile and work in Dialog ' s system, an d used it as the core of VANF. Currently, about 30% of VANF rules are derived from FASTUS. Still, VANF should be considered a variation on the theme of FASTUS. Two thousand documents from different newspapers were processed by human content specialists and th e names to be extracted as companies were tagged. The definition of a \"company\" was \"user oriented \" (e .g. football teams were to be considered companies, while publications, trade unions and governmen t organizations were not). One thousand of the documents were then used as a training set, and the othe r 1000 as a blind test set .", "pdf_parse": { "paper_id": "M95-1008", "_pdf_hash": "", "abstract": [ { "text": "Knight-Ridder Information, Inc., participated in MUC-6 with VANF (Valu e Adding Name Finder), the system used by Knight-Ridder Information i n production for adding a Company Names descriptor field to online newspape r and newswire databases. Knight-Ridder Information participated in the NE tas k only. The system used for MUC-6 is exactly the same as is used in production. The only difference is the input and output formats. VANF used a cascaded non-deterministic state machines approach and is based on FASTUS. 1. Backgroun d Late 1992, we realized that a company like Knight-Ridder Information (Dialog Information Services, a t that time) had to get into the business of Value Added Reselling of information. An obvious question was: what kind of value can be added to the information? The obvious answer was that the added value shoul d have the same nature as the original : it should be information. The plan was to introduce NLP into Dialog's technology to add informational value to raw documents. The industrial environment required that anything we work on has a fmite development time and is usable in production after the developmen t is over. This dictated step-wise development and required starting with solvable problems. The obvious choice was smart tokenization, which included named entity detection. Originally, we tried to buy the technology from outside, so that a full-fledged partnership could b e developed based on the initial project. Two first attempts ended in nothing (although they were quite useful experiences for Dialog, and probably for the potential partners as well). The third one, with SRI , led to development of the current system, VANF (Value Adding Name Finder), which is routinely used i n production at Knight-Ridder Information. VANF is an independent re-implementation of FASTUS [I]. At first, we wanted to use FASTUS as it was. However, the production requirements prohibited using a Lisp environment, and we decided to port it (to c++). A direct port was not feasible ; at the same time, SRI decided that a declarative, grammar-like description of FSM was more intuitive and easier to work wit h than the graphical tools they were using at the time. So a grammar definition language was defined, w e spent some time discussing and refining it, and I implemented an engine for Dialog's version of the language. Under the contract we had, SRI delivered a transcription of FASTUS rules in the new (declarative) language, although they did not have their own interpreter (nor even a complete languag e definition) at that time. Somehow, we made this transcription compile and work in Dialog ' s system, an d used it as the core of VANF. Currently, about 30% of VANF rules are derived from FASTUS. Still, VANF should be considered a variation on the theme of FASTUS. Two thousand documents from different newspapers were processed by human content specialists and th e names to be extracted as companies were tagged. The definition of a \"company\" was \"user oriented \" (e .g. football teams were to be considered companies, while publications, trade unions and governmen t organizations were not). One thousand of the documents were then used as a training set, and the othe r 1000 as a blind test set .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Abstract", "sec_num": null } ], "body_text": [ { "text": "The overall effort on VANF was about 8 months of one developer' s time (developing the engine, the lexicons, and the rules), 6 person-months of the content specialists' effort, and about 6 person-months o f project management effort.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null }, { "text": "System Descriptio n A grammar defines a set of NDFSM and the sequence of their application.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "The terminal symbols in the grammar are literal words and combinations of syntactic and dictionary flags .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "Given a NDFSM SKand a sequence of symbols .SS \u2022 all the paths are followed;", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "\u2022 the longest matching sequence 3t1 ..i] is considered the result of the application;", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "\u2022 the actions corresponding to all the longest paths are executed ;", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "\u2022 a single output symbol (i .e . a head word with a bunch of attributes) is sent to the next stage .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "Then 9Kis applied to the next segment o f", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "The output of his used as the input for another NDFSM . Each stage is responsible for specific kind of processing (see below) .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "This approach allows us to restrict non-determinism and prevent combinatorial explosion. The expressive power of our formalism is the same as that of the attribute grammars ; at the same time the efficiency i s close to that of finite state machines .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "The approach is very similar to that used in FASTUS [1] .", "cite_spans": [ { "start": 52, "end": 55, "text": "[1]", "ref_id": "BIBREF0" } ], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "Evidence combiner and knowledge base . We are not interested in the NDFSM output per se . The pattern matching results are collected from the side-effect of the actions attached to the rules.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "Most important are the actions which assert facts extracted from the text (in the VANF case, the facts are in the form \"NNN is a name of an entity of type X\") . These assertions are not treated as the ultimate truth , but all the evidence is collected and combined by an evidence combiner .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": ".", "sec_num": "2" }, { "text": "\u2022 resolving conflicts between contradicting evidenc e checking the names against the known names database", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This include s \u2022 matching the variants of names", "sec_num": null }, { "text": "The document scanner parses the input documents in their native format (fields, SGML tags, etc .) . breaks a document into paragraphs, and does the lexicon look-up . It also takes care of entities like simpl e dates, phone numbers, and other regular quasi-lexical units . This part of the scanner is based on regular expressions (lex) .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "This include s \u2022 matching the variants of names", "sec_num": null }, { "text": "1) The basic tokenizer (written in lex) handles lexicon lookup, capitalization, simple money, number , phone-number, and date expression processing, simple unknown names processing : multi-word lexicon entries are handled here as well .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Text processing flow", "sec_num": "2.2." }, { "text": "2) The preprocessor (1st stage of NDFSM) looks for sentence boundaries, more complicated number an d date expressions, obvious names type determination and other names bracketing .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Text processing flow", "sec_num": "2.2." }, { "text": "3) The chunker (2nd stage of NDFSM) brackets noun phrases and verb groups, replacing them with thei r head words . 4) The pattern matchers; several stages of NDFSM detect certain interesting types of noun groups using patterns( \"ABC 's president X\", and look for other patterns relevant to the VANF task; for each interesting pattern an attached function is executed to pass on a , candidate name to the evidenc e combiner. 5) The evidence combiner (C++) merges similar names and assigns types to the names whose type could not be derived from the form or from the context ; database lookup happens here as well . ", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Text processing flow", "sec_num": "2.2." }, { "text": "Knight-Ridder Information, Inc . did only the NE part.", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\"Where the bulk of effort was spent \"", "sec_num": "3.1." }, { "text": "The original VANF development tool : 4 months -the engine 4 months -the rules and improving the lexico n Then, for the MUC-6 evaluation : 1 month -preparation for MUC ; this included improving the rules based on the training corpu s and making the system interface component to produce output, which made the MU C scorer happy .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "\"Where the bulk of effort was spent \"", "sec_num": "3.1." }, { "text": "Processing rate : 5 Mb / hour Reloading the lexicon : 20 se c Reloading the grammar: 20 sec 3.3. Training and rules modificatio n All rules development was done by hand . The engine provides a lot of tracing information and allows easy definition of which rule(s) contributed to a specific decision . This should help to implement automati c training (see below) in the future . At present, this allows us to modify rules . Also, the reason we could ge t anywhere at all was the fast turn-around cycle : modifying the grammar, and rerunning a document take s about 1 minute .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Technical characteristics", "sec_num": "3.2." }, { "text": "Our experience with VANF has proved that a core cascaded NDFSM approach is suitable for man y intelligent text processing tasks . As it was pointed out in [1] , an efficient implementation and easy-to-us e tools allowed us to tune a pretty primitive technology to produce quite useful results . In our case, suc h efficiency made it possible to build a rule set consisting of many quite specific rules, such that althoug h each one has a limited application, together, they cover a large area . In the future, such rule sets should b e constructed using automated tools . Two future tasks will be concentrated on :", "cite_spans": [ { "start": 155, "end": 158, "text": "[1]", "ref_id": "BIBREF0" } ], "ref_spans": [], "eq_spans": [], "section": "Experience gained from VANF and the futur e", "sec_num": "3.4." }, { "text": "\u2022 training / rules building tools ; the author plans to develop a learn-by-example system, conceptuall y similar the Autoslog [2] and the like ; \u2022 non-boolean evidence combining .", "cite_spans": [ { "start": 126, "end": 129, "text": "[2]", "ref_id": "BIBREF1" } ], "ref_spans": [], "eq_spans": [], "section": "Experience gained from VANF and the futur e", "sec_num": "3.4." }, { "text": "The lexicon contains word forms with flags ; the flags encode POS and semantic information . There is no formal differentiation between primary (N, V) and secondary (Sing, Trans) morphological features, no r between grammatical and semantic flags . The flags corresponding to different meanings of a word are mixed in the same entry . ( Aux-Be --> { Bel I Be-Not I Aux-Modal 'be' I Aux-Have 'been ' } ('being ' ) ; \u2022 Bel --> { 'am' I 'is' << Sing = T >>l 'are'", "cite_spans": [], "ref_spans": [ { "start": 335, "end": 336, "text": "(", "ref_id": null } ], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "''re' I 'was' << Sing = T >>'were '", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "Be-Not }-> { Bel 'not' I 'ain't' I 'aren ' t ' I 'isn't' << Sing = T >> I 'wasn't ' < < Sing = T >> I 'weren't \"", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "; neg = T", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "The following rule belongs to a sub-NDFSM which consumes the output of the \"syntax phase\" and detect s person names in contexts like \"Neither of Makoto Suzuki's parents\" or \"Mary was Joe's wife\" :", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "Name-In-Context --> (Possible-Person-Name {',' I Be}) { Possible-Person-Name 1 N G } \u2022's' NG[-Indef,$relative$]", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": ". .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "Appendix 1 : Lexicon and Grammar Sample s", "sec_num": "4." }, { "text": "5.1 . Errors list. 2) \"Interpublic Group's McCann-Erickson\" and \"WPP Group's J. Walter Thompson\" were treated as single entities ; this is strange, because we have rules to split them, and the evidence combiner mus t have picked the second parts of these names, because they occur elsewhere . A possible explanation i s that Dialog' s original requirements explicitly advised against splitting the names, which might mak e sense in the first of the two examples. 3) \"60 pounds\" taken for a monetary expression . 4) All occurrences of \" Coke\" were ignored. 5) \"New York Times\" ignored -according to Dialog's spec. 6) The Title field was not processed at all -for no good reason . 7) \"Other ad agencies, such as Fallon McElligott,\" -a bu g Asked why he would choose to voluntarily exit while he still i s so young, Mr . James, who has a reputation a s an extraordinarily tough taskmaster, says that because he 'had a great time' in advertising,' he doesn't want to 'talk about th e disappointments .' In fact, when he is asked his opinion of the new batch of Coke ads from CAA, Mr . James places his hands over hi s mouth . He shrugs . He doesn't utter a word . He has, he says, fon d memories of working with Coke executives . 'Coke has given us great highs,' says Mr . James, sitting in his plush office , filled wit h photographs of sailing as well as huge models of, among othe r things, a Dutch tugboat .", "cite_spans": [ { "start": 19, "end": 21, "text": "2)", "ref_id": "BIBREF1" } ], "ref_spans": [], "eq_spans": [], "section": "Appendix . 2. The Walk-Through Exampl e", "sec_num": "5." }, { "text": "He says he feels a 'great sense of accomplishment .' In 3 6 countries, McCann is ranked in the top three ; in 75 countries, it i s in the top 10 .

Soon, Mr . James will be able to compete in as man y sailing race s as he chooses . And concentrate on his duties as rear commodore a t the New York Yacht Club .

Maybe he'll even leave something from his office for Mr . Dooner . Perhaps a framed page from the New York Times , dated Dec . 8, 1987 , showing a year-end chart of the stock market crash earlier tha t year . Mr . sENAMEX TYPE=\"PERSON'>James says he framed it and kept it by hi s desk as a 'personal reminder . It can all be gone like that . ' 6 .", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "

", "sec_num": null }, { "text": "Currently with Verity, Inc . reachable as arkady@verity . cor n", "cite_spans": [], "ref_spans": [], "eq_spans": [], "section": "", "sec_num": null } ], "back_matter": [], "bib_entries": { "BIBREF0": { "ref_id": "b0", "title": "FASTUS : A Finite-state Processor fo r Information Extraction from Real-world Text", "authors": [ { "first": "Douglas", "middle": [ "E" ], "last": "Appelt", "suffix": "" }, { "first": "Jerry", "middle": [ "R" ], "last": "Hobbs", "suffix": "" }, { "first": "David", "middle": [ "Israel" ], "last": "Mabry Tyson", "suffix": "" } ], "year": null, "venue": "IJCAI 93, 13 Joint Conference of Artificia l Intelligence", "volume": "", "issue": "", "pages": "", "other_ids": {}, "num": null, "urls": [], "raw_text": "Douglas E . Appelt, Jerry R . Hobbs , David Israel. Mabry Tyson \"FASTUS : A Finite-state Processor fo r Information Extraction from Real-world Text\" . In IJCAI 93, 13 Joint Conference of Artificia l Intelligence .", "links": null }, "BIBREF1": { "ref_id": "b1", "title": "Automatically Constructing a Dictionary for Information Extraction Tasks", "authors": [ { "first": "E", "middle": [], "last": "Riloff", "suffix": "" } ], "year": null, "venue": "Proceedings of the 11th National Conference on Artificial Intelligence", "volume": "", "issue": "", "pages": "811--816", "other_ids": {}, "num": null, "urls": [], "raw_text": "Riloff, E. \"Automatically Constructing a Dictionary for Information Extraction Tasks \" Proceedings of the 11th National Conference on Artificial Intelligence, 811-816 .", "links": null } }, "ref_entries": { "TABREF2": { "content": "
aging N Sing V-Ing Tran s
agitate V Trans
agitated V-Ed-En Trans
agitates V-S Tran s
Argentine Adj Noun-Like
Argentine N Sing Adjnou n
Argentinean N Sing Adjnoun NatiOnality
Argentineans N-S Nationalit y
Arizona N Sing Country Stat e
'so called' Adj
'so far' Adv
'so that' Subcon j
'soft drink' N Sing
'soft drinks* N-S
'software house' N Sing
$relative$ -->
'mother' 'father '
'parents' \"grand-mother' 'grand-father' 'grand-parents '
'son''sons''daughter' 'daughters'\u2022child '
'children' 'grand-daughter' 'grand-son '
'wife' 'ex-wife' 'husband' 'ex-husband' \u2022spouse \u2022
'sister' 'sisters' 'brother' *brothers' 'sibling' 'siblings '
4.2VG not been
Aux-Be-VG --> {Ax-Be
Aux-Modal Adv* 'be '
Aux-Have 'been '
}
t \"is [broken]' 'isn't [really broken]', 'will often be [broken]' , % 'can't have been [broken]', 'has been [broken)\", 'had not been [broken] . \"
", "html": null, "num": null, "text": "Therefore, in the grammar, N[Country] or Country[N] are equivalent notations . One can also write N[V] meaning \" any word which is both a verb and a noun\" . )4.1 . A sample of the lexiconLexical information can be also stored in the grammar description in the form of word lists .For example, the \"syntax stage \" of the VANF NDFSM contains rules :", "type_str": "table" }, "TABREF3": { "content": "
</p > Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> met with <ENAMEX TYPE='PERSON'>Marti n <p> Puris</ENAMEX>, president and chief executiv e
officer of <ENAMEX TYPE='ORGANIZATION\">Ammirati & Puris</ENAMEX>, about <ENAME X Even <ENAMEX TYPE='</p > <p> While McCann's world-wide billings rose <NUMEX TYPE='PERCENT'>12%</NUMEX> t o <NUMEX TYPE='MONEY'>$6 .4 billion</NUMEX> las t TYPE='</p>
<p>
Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> is just gearing up for the headaches o f
running one o f </p > the largest world-wide agencies . (There are no immediate plans t o
<p> replace Mr . <ENAMEX TYPE = \" PERSON'>Dooner</ENAMEX> as president ; Mr . <ENAME X <ENAMEX TYPE='ORGANIZATION'>McCann</ENAMEX> still handles promotions and medi a TYPE='PERSON'>James</ENAMEX> operated as chairman , buying for Coke . But chief executive officer and president for a period of time .) Mr .
the bragging rights to Coke's ubiquitous advertising belongs t o <ENAMEX TYPE='PERSON\">James</ENAMEX> is filled with thoughts of enjoying his thre e
5 .2 . hobbies : The output sailing, skiing and hunting . <ENAMEX TYPE='striving to have a strong renewed creative partnership with </p> <DOCID> wsj94_026 .0231 </DOCID> Marketing & Media --Advertising : other ad agencies, such as <ENAMEX TYPE='PERSON\">Fallon McElligott</ENAMEX>, wil l @ At Helm of McCann-Erickso n is that <ENAMEX TYPE='ORGANIZATION'>CAA</ENAMEX> and @ John Dooner Will Succeed Jame s slim since word from Coke headquarters in <ENAMEX TYPE='LOCATION'>Atlanta</ENAMEX > <HL> happening are <DOCNO> 940224-0133 . </DOCNO> Coca-Cola,' Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> says . However, odds of tha t <p>
continue t o @ By Kevin Goldman </HL > handle Coke advertising . <DD> <TIMEX TYPE='DATE'>02/24/94</TIMEX> </DD > </p > <SO> WALL STREET JOURNAL (J), PAGE B8 </SO > <CO> <p> IPG K </CO > Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX>, who recently lost <NUME X <IN> ADVERTISING (ADV), ALL ENTERTAINMENT & LEISURE (ENT) , TYPE='MONEY'>60 pounds</NUMEX> over three-and-a-hal f FOOD PRODUCTS (FOD), FOOD PRODUCERS, EXCLUDING FISHING (OFP) , months, says now that he has 'reinvented' himself, he wants to d o
RECREATIONAL PRODUCTS & SERVICES (REC), TOYS (TMF) </IN> the same for the agency . For Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX>, it mean s <TXT > <p> maintaining hi s
running and exercise schedule, and for the agency, it mean s One of the many differences between <ENAMEX TYPE='PERSON'>Robert L . developing more global campaigns that nonetheless reflect loca l James</ENAMEX>, chairman and cultures . One <ENAMEX TYPE='ORGANIZATION'>McCann</ENAMEX> account, 'I Can't Believ e chief executive officer of <ENAMEX TYPE='ORGANIZATION'>McCann-Erickson</ENAMEX>, and It's Not Butter,' a <ENAMEX TYPE='PERSON'>John J . Dooner Jr</ENAMEX> . , butter substitute, is in 11 countries, for example . the agency's president and chief operating officer, is quit e < / p > telling : Mr . <ENAMEX TYPE = ' PERSON'>James</ENAMEX> enjoys sailboating, while Mr . <p> <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> owns a <ENAMEX TYPE='ORGANIZATION'>McCann</ENAMEX> has initiated a new so-called global powerboat . collaborative system ,
</p> composed of world-wide account directors paired with creativ e <p> partners . In addition, <ENAMEX TYPE='PERSON'>Peter Kim</ENAMEX> was hired fro m Now, Mr . <ENAMEX TYPE= ' PERSON'>James</ENAMEX> is preparing to sail into th e <ENAMEX TYPE='ORGANIZATION'>WPP Group's J . sunset, and Mr . Walter Thompson</ENAMEX> last <TIMEX TYPE='DATE'>September</TIMEX> as vice chairman , <ENAMEX TYPE = ' PERSON'>Dooner</ENAMEX> is poised to rev up the engines to guid e chief strategy <ENAMEX TYPE='ORGANIZATION'>Interpublic Group' s officer, world-wide . McCann-Erickson</ENAMEX> into the 21st century . Yesterday, <ENAME X </p >
TYPE=' ORGANIZATION'>McCann</ENAMEX> mad e <p> official what had been widely anticipated : Mr . <ENAMEX TYPE='PERSON'>James</ENAMEX> , Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> doesn't see a creative malais e 57 years old , permeating the agency . is stepping down as chief executive officer on <TIMEX TYPE='DATE'>July 1</TIMEX> an d He points to several campaigns with pride, including the Taster' s wil l Choice commercials that are like a running soap opera . 'It's a <NUMEX
retire as chairman at the end of the year . He will be succeeded by TYPE='MONEY'>$1 9 Mr . <ENAMEX TYPE=' PERSON'>Dooner</ENAMEX>, 45 . </NUMEX>million campaign with the recognition of a <NUMEX TYPE='MONEY'>$20 0 </p > million</NUMEX> campaign, ' <o> he says of the commercials that feature a couple that must hold a It promises to ba a smooth process, which is unusual given th e record for the length of time dating before kissing .
volatile atmosphere of the advertising business But Mr . <ENAMEX </p > TYPE='PERSON'>Dooner</ENAMEX> has <p>
a big challenge that will be his top priority . 'I'm going to focu s Even so, Mr . <ENAMEX TYPE='PERSON'>Dooner</ENAMEX> is on the prowl for mor e
on strengthening the creative work,' he says . 'There is room t o creative talent an d grow . We can make further improvements in terms of the perception o f is interested in acquiring a hot agency . He says he would like t o
our creative work .' finalize an acquisition 'yesterday . I'm not known for patience . '
</p>
<p>
", "html": null, "num": null, "text": "PERSON'>Alan Gottesman, an analyst with PaineWebber, who believe s McCann is filled with 'vitality' and is i n 'great shape,' says tha t from a creative standpoint, 'You wouldn't pay to see their reel' o f commercials . of the key creative assignment for the prestigious Coca-Col a Classic account . 'I would be less than honest to say I'm no t disappointed not to be able to claim creative leadership for Coke, ' Mr . Dooner says . ORGANIZATION'>McCann's acquiring the agenc y with billings of $400 million, but nothing ha s materialized . 'There is no question,' says Mr . Dooner, 'that we are looking for qualit y acquisitions and Ammirati & Puris is a qualit y operation . There are some people and entire agencies that I would love to see be part o f the McCann family .' Mr . Dooner declines to identify possibl e acquisitions .", "type_str": "table" } } } }