{
    "paper_id": "J75-4023",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T02:40:39.519906Z"
    },
    "title": "! 3 7 5 ACL Meeting SYNTACTIC PROCESSING IN THE BBN SPEECH UNDERSTANDING SYSTEM",
    "authors": [
        {
            "first": "M",
            "middle": [],
            "last": "A D E L I N E Bates",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "Bolt Belranek a n d Newman Inc",
                "location": {}
            },
            "email": ""
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "",
    "pdf_parse": {
        "paper_id": "J75-4023",
        "_pdf_hash": "",
        "abstract": [],
        "body_text": [
            {
                "text": "The syntactic analysis system presented h e r e is composed of ~v o p a r t s , a m o d i f i e d augmented rransition network grammar and a 7 a r s e r which i s designed f o r a speech understanding environment.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            },
            {
                "text": "The parser o p e r a t e s on p a r t i a l u t t e r a n c e s c a l l e d t h & r i e s . ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [],
        "bib_entries": {},
        "ref_entries": {
            "FIGREF0": {
                "num": null,
                "text": "t 3 e o r y may be thought of as a s e t of words which a r e hypothe? s i z e d to be in t h e uttezance. The p a r s e r processes t h e words in a theory by building partial syntactic paths using t h e words of :he t h e o r y . These paths do not depend on left c o n t e x t , which w i l l Le missing i f t h e r e a r e gaps i n t h e t h e o r y . S y n t a c t -i c constitu-e x s a r e built where p o s s i b l e and, whenever a constituent is i t the parser can interface witrh the semantic zomponent of he t o t a l . s~e e c h understanditzg system f o r guidance and v e r i f i c at i o n . The parser t r i e s to p r e d i c t words a n d / o r syntactic c a t e g o r i e s 50 f i l l o r reduce g a p s i n the t h e o r y , particulatly small function r i o r l s which a r e difficult to d e t e c t r e l i a b l y on acoustic grounds aloone. Thk p a r s e r does not follow a l l possible parse p a t h s , but n c t c~; :~p t s t o select the.nost likely ones fo; extension. It uses a '-:dicious mixture o f t o p down, bottom up, depth first, and breadth J \" I r l r s t parsing strategies t o take advantage of local, r e l i a b l e in-forrLstion. It saves all t h e information gained while following a l~c r n a t i v e p a r s e paths, so t h a t s e v e r a l p a r s e p a t h s which share a comnon p a r t , even i f t h e p a t h s a r e in different t h e o r i e s , can s t a r e c h a t p o r t i o n without ~e p a r s i n g . This i s true even if the -. r s e r/ ' .A p a t h s s p l i t b e f o r e a n d / o r a f t e r t h e common part and even if t h e common s e c t i o n analyzes only p a r t of a syntactic constituent.",
                "uris": null,
                "type_str": "figure"
            }
        }
    }
}