Michael J. Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics, Santa Cruz, Cal., pages 184--191, 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....conditional probability context free grammar from an unlabeled bracketed corpus based on clustering analysis, and introduces a natural language parsing model which uses a probability based scoring function of the grammar to rank parses of a sentence. The method is superior to previous works (i.e. [Collins, 1996]) in the following points: rst the method can deal with training corpora with no nonterminal labels. Second, unlike inside outside algorithm based methods, the acquired grammars are not restricted to Chomsly normalform CFGs and less computational cost for training is required. Third, the learned ....

Michael John Collins. A new statistical parser based on bigram lexical dependencies. In Proc. of the 34th ACL, pages 184-191, 1996.

....has been devoted to the induction of context free grammars. CFGs can be used for processing natural languages, too. Automatic induction of a natural language grammar by only using example sentences draws attention. Although statistical approaches form the main stream among the researchers [7, 6, 8], several attempts have been carried out to attack the problem with evolutionary techniques, too. For instance, 55] proposes a method based on genetic algorithms. The authors claim that although statistical methods o er a possible solution to the problem, drawbacks exist. It is quite dicult to ....

Michael John Collins. A new statistical parser based on bigram lexical dependencies. In Arivind Joshi and Martha Palmer, editors, Proceedings of the ThirtyFourth Annual Meeting of the Association for Computational Linguistics, pages 184-191, San Francisco, 1996. Morgan Kaufmann Publishers.

....bene t from shallow parsing. But for Natural Language Understanding, we claim that a parser should meet these traditional requirements. Researchers have had much success in recent years in the automatic reconstruction of parse trees of the sort annotated in the Penn Treebank [Magerman, 1995; Collins, 1996; Charniak, 1997a; Ratnaparkhi, 1997] But do these trees readily admit semantic interpretation The trees tend to have a very at structure; base NPs are famously left unstructured. Section 00 alone of the Penn Treebank s annotation of the Wall Street Journal corpus has 730 di erent expansions ....

.... almost all research using the Penn Treebank to some extent incorporates linguistic assumptions into the work: deciding how to decompose a parse tree into its constituent pieces presumes some linguistic underpinnings (e.g. Charniak [1997a] supposes they are licensed by a context free grammar, and Collins [1996; 1997] instead treats them as sequences of head daughter attachment decisions) Any lexicalization of the treebank requires some assumption about the headedness of structures; most researchers use some variant of the rules constructed by Magerman [1995] These rules designate a single word in any ....

Michael Collins, \A New Statistical Parser Based on Bigram Lexical Dependencies, " In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics, pages 184-191, 1996.

.... Joakim Nivre 1 Introduction One of the most prominent trends in research on natural language parsing during the last decade has been the development of data driven methods and in particular the use of stochastic parsing models (see, e.g. Magerman [18] Bod [2, 3] Charniak [6, 7] Collins [8, 9, 10] and Goodman [13] Another trend, although perhaps less prominent, has been a growing interest in models based on dependency grammar, exempli ed by the collection of papers in Kahane and Polgure [17] In this paper, I will be concerned with the combination of stochastic parsing and dependency ....

Michael Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annatual Meeting of the Association for Computational Linguistics, pages 184191, Santa Cruz, CA, 1996.

....a long and venerable tradition in linguistics, dependency grammar has until quite recently played a fairly marginal role in natural language parsing. However, dependency based representations have turned out to be useful in statistical approaches to parsing and disambiguation (see, e.g. Collins [4, 5, 6], Eisner [15, 16, 17] Samuelsson [25] and they also appear well suited for languages with less rigid word order constraints (Covington [9, 10] Collins et al. 7] Several di erent parsing techniques have been used with dependency grammar. The most common approach is probably to use some ....

Michael Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annatual Meeting of the Association for Computational Linguistics, pages 184191, Santa Cruz, CA, 1996.

.... a complete analysis of a set of selected texts for each query and of the query itself and creates, after several intermediate steps, a logical representation inspired by the notation proposed by Hobbs (on which we also base our MLFs) The syntax analysis in Falcon is based on a statistical parser [4] while we use a dependency parser that computes all syntactically possible structures which we then filter according to a combination of hand crafted rules and Brill and Resnik disambiguation procedure [2] A similarity between ExtrAns and Falcon is that both build a semantic form starting from a ....

Michael Collins, `A new statistical parser based on bigram lexical dependencies ', in Proceedings of the 34st Annual Meeting of the Association for Computational Linguistics, ACL-96, pp. 184--191, (1996).

....6.6 Correlation Between Normalized Number of Open Class Heads and Difficulty . 6.7 Slight Correlation Between Normalized Number of Open Class Bigrams and Difficulty . viii Chapter I Introduction Statistical parsers ( 13] [8, 7], 6] have the advantage of being able to learn models directly from labeled data and to rank multiple parses so that there is some criterion for choosing among them. However, because of the large amount of computation required and the ambiguity of the grammars used, statistical parsers are ....

Michael John Collins. A new statistical parser based on bigram lexical depen- dencies. In Proceedings of the 3,ith Annual Meeting of the Association for Com- putational Linguistics, pages 177-183, 1996.

.... (MacDonald 94; Jurafsky 96; Ratnaparkhi 97) On the one hand, the preferences are stated according to structure based heuristics (Walther 93) on the other hand they are stated according to probabilities estimated by corpus based frequencies (Magerman Marcus 91; Briscoe Carroll 93; Bod 95; Collins 96) Until recently, symbolic pruning devices were used with symbolic parsers rich structural descriptions generated according to a symbolic grammar while statistical pruning devices were used with statistical parsers raw descriptions generated according to a statistical language model. A ....

....of the probability model is that the event space is defined at two levels of granularity lexical if the verb is involved, syntactic for all other relations. Thus this model is fundamentally different from other dependency based statistical parsers which are completely lexicalized, such as (Collins 96; Charniak 95) If the site x j is a VP and the dependency dep is the subcategorized complement attachment, this means that the base x i is an argument of the verb. Then, these probabilities are estimated by taking into account lexical elements (the verb) We determine the global distribution of ....

Michael Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the Association for Computational Linguistics, Santa Cruz, California, 1996.

....and Binarization 14 percolate headwords using a context free (CF) rule based approach and then binarize the parses by again using a rule based approach. Headword Percolation Inherently a heuristic process, we were satisfied with the output of an enhanced version of the procedure described in [11] also known under the name Magerman Black Headword Percolation Rules . The procedure first decomposes a parse tree from the treebank into its contextfree constituents, identified solely by the non terminal POS labels. Within each constituent we then identify the headword position and then, ....

....assumptions whichmay not be and probably are not correct and thus have a damaging effect on the modeling power of our model. The equivalence classification should identify the strong predictors in the context and allow reliable estimates from a treebank. Our choice is inspired by [11] and intuitively explained in Section 2.2: P (w k jW k;1 T k;1 ) P (w k j[W k;1 T k;1 ] P (w k jh 0 #h ;1 ) 2.3) P (t k jw k #W k;1 T k;1 ) P (t k jw k # [W k;1 T k;1 ] P (t k jw k #h 0 :tag# h;1 :tag) 2.4) P (p k i jW k T k ) P (p k i j[W k T k ] P (p k i jh 0 #h ;1 ) 2.5) ....

[Article contains additional citation context not shown here]

Michael John Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics, pages 184--191. Santa Cruz, CA, 1996.

....In addition to glossary look up, the lexical analyzer uses rules to interpret dates and numerical expressions such as 1.2 x 1.7 cm . The syntactic parser creates a dependency graph with an arc from each word to the word it modifies. We use a statistical parser somewhat similar to that of Collins [1996] and of Eisner [1996] in which the probability of an arc between each pair of words is estimated from hand tagged training sentences. Dynamic programming is used to compute the parse graph with maximum joint probability, subject to some constraints on well formedness of the graph. The next step ....

Michael Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34 th Annual Meeting of the Association for Computational Linguistics, pages 184-191, 1996.

....structure to better extract plausible answers from the pages returned by the search engine. Natural language parsing is a mature eld with a decades long history,sowechose to adopt the best existing parsing technology rather than roll our own. Today s best parsers employ statistical techniques [12, 8]. These parsers are trained on a tagged corpus, and they 151 user Search Engine Question Classification Link Parser Classification Rules Query Formulation Query Formulation Rules (only rules applicable to the question will be triggered) WordNet Transformational Grammar ....

....entity s role in the sentence, and a governing word to which the entity stands in the semantic relation. Another system, described by Harabagiu [15] achieves respectable results for ###### by reasoning with linkages between words, which are obtained from its dependency based statistical parser [12]. Harabagiu s system rst parses the question and the retrieved documents into word linkages. It then transforms these linkages into logical forms. The answer is selected using an inference procedure, supplemented by an external knowledge base. We believe mechanisms of this type would be very ....

Michael John Collins. A New Statistical Parser Based on Bigram Lexical Dependencies. In 160 Proceedings of the 34th Annual Meeting of the ACL, Santa Cruz, 1996.

....as q n (this is only approximate because the themselves were derived from the sentence . Sometime is further approximated as , where is the single best scoring parse. An example of such a model is [74] which uses the parser of [75] to generate the candidate parses, and trains the parameters using maximum entropy. The probabilistic link grammar [44] mentioned in section 3.3 also falls roughly in this category. Most recently, 76] employed a parser with probabilistic parameterization of a pushdown automata, and used an ....

Michael Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th annual meeting of the association for Computational Linguistics, pages 184--191, May 1996.

No context found.

Michael J. Collins. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics, Santa Cruz, Cal., pages 184--191, 1996.

No context found.

Michael Collins. 1996. A new statistical parser based on bigram lexical dependencies. Proceedings of the 34th Meeting of the Association for Computational Linguistics (pp. 184-191). Santa Cruz, CA.

No context found.

Michael Collins. 1996. A new statistical parser based on bigram lexical dependencies. In Proc. of ACL-96.

No context found.

Michael J. Collins. A New Statistical Parser Based on Bigram Lexical Dependencies. In Proceedings of the 34th Annual Meeting of the Association for Computational Linguistics (ACL), Santa Cruz, USA, 1996.

No context found.

Michael John Collins. A new statistical parser based on bigram lexical dependencies. In Proc. of the 34th Annual Meeting of the ACL, pages 184-- 191, 1996.

No context found.

Michael John Collins. A new statistical parser based on bigram lexical dependencies. In ACL 34, pages 184-- 191, 1996.

No context found.

Michael John Collins, "A new statistical parser based on bigram lexical dependencies", in: Proceedings of the 34rd Annual Meeting of the Association for Computational Linguistics, 1996.

No context found.

Michael John Collins, "A new statistical parser based on bigram lexical dependencies", in: Proceedings of the 34rd Annual Meeting of the Association for Computational Linguistics, 1996.

No context found.

Michael Collins. 1996. "A New Statistical Parser Based on Bigram Lexical Dependencies". In Procs. 34th Annual Meeting of the Associa- tion for Computational Linguistics, pp. 184-191. ACL.

No context found.

Michael John Collins (1996). "A New Statistical Parser Based on Bigram Lexical Dependencies." In Proc. of ACL-34, 184-191.

No context found.

Michael Collins. 1996. "A New Statistical Parser Based on Bigram Lexical Dependencies". In Procs. 34th Annual Meeting of the Association for Computational Linguistics , pp. 184--191. ACL.

No context found.

Michael Collins. 1996. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 34th Annual Meeting of the ACL.

No context found.

Michael Collins. 1996. A new statistical parser based on bigram lexical dependencies. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics, Santa Cruz, California.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC