Hermjakob, U., & Mooney, R. (1997). Learning Parse and Translation Decisions From Examples With Rich Context. Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics (ACL/EACL97) .  Home/Search   Document Details and Download   Summary   Related Articles   Check

....it from others [7] 1.b to facilitate fallback (query generalization) remove from the signatures all words or phrases that occur in fewer than half the texts of the topic group 1.c save the signatures in a tree, organized by signature overlap. We use the format of Webclopedia s parser CONTEX [5,6]; see Figure 1 1.d use Webclopedia s ranking algorithm to rank sentences [8] 3.2 Filter for Content Given the ranked list of sentences, re rank or remove those according to the following conditions: 2.a remove all sentences with sentence position 10. This is a simple version of SUMMARIST s ....

Hermjakob, U. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. Ph.D. dissertation, University of Texas at Austin. file://ftp.cs.utexas.edu/pub/~mooney/papers/hermjakobdissertation -97.ps.gz.

....representation, such as the boy NP[the,boy] step 3) Phrase labels such as NP and RC are only used in this figure to make the process clear. The general SR model can be implemented in many ways. A set of symbolic shift reduce rules can be written by hand or learned from input examples [Hermjacob and Mooney, 1997; Simmons and Yu, 1991; Zelle and Mooney, 1996] It is also possible to train a neural network to make parsing decisions based on the current stack and the input buffer. If trained properly, the neural network can generalize well to new sentences [Simmons and Yu, 1992] Whatever correlations ....

Ulf Hermjacob and Raymond J. Mooney. Learning parse and translation decisions from examples with rich context. In Proceedings of the 35th Annual Meeting of the ACL, 1997.

....producing annotated treebanks is often prohibitive, thus rendering manual construction of such data for new languages infeasible. Some researchers have developed techniques for fast acquisition of hand annotated Treebanks [7] Others have developed machine learning techniques for inducing parsers [9, 10], but these require extensive collections of complex translation pairs for broadscale MT. Because divergences generally require a combination of lexical and structural manipulations, they are handled traditionally through the use of transfer rules [12, 13] Unfortunately, automatic extraction of ....

Hermjakob, U., Mooney, R.J.: Learning Parse and Translation Decisions from Examples with Rich Context. In: Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics and Eighth Conference of the European Chapter of the Association for Computational Linguistics. (1997) 482-489

.... context free grammars, and produce more accurate results if they learn about bilexical dependencies between head words of constituents [2, 3] Though in general these approaches use specialized machine learning techniques, general learning frameworks are also applicable (ID3 algorithm [7], maximum entropy model [14] The most common control structure is the chart based (or dynamic programming) technique. This paper explores the possibility that a psycholinguistically motivated parser can also perform well on freely occurring text. Our reference theories in the psycholinguistic ....

U. Hermjakob and R. J. Mooney. Learning parse and translation decisions from examples with rich context. In Proceedings of ACL97, pages 482-489, 1997.

.... no explicit rules, such as neural networks (e.g. Buo 1996] or approaches where the machine learning algorithms attempt to infer via deduction (e.g. Samuelsson 1994] induction (e.g. Theeramunkong et al. 1997] Zelle Mooney 1994] under user cooperation (e.g. Simmons Yu 1992] [Hermjakob Mooney 1997]) transformation based error driven learning (e.g. Brill 1993] or even decision trees (e.g. Magerman 1995] a grammar from raw or preprocessed data. In our work, we do not wish to acquire a grammar: we have one and want to devise a mechanism to make some of its parts adaptable to the corpus ....

Hermjakob U. & Mooney R.J. (1997) "Learning Parse and Translation Decisions From Examples With Rich Context", Proc. of ACL-EACL Conf., pp.482-489.

....(1995) also train history based decision tree models from a treebank for use in a parser, but do not require an explicit hand written grammar. These decision trees do not look at words directly, but instead represent words as bitstrings derived from an automatic clustering technique. In contrast, Hermjakob and Mooney (1997) use a rich semantic representation when training decision tree and decision list techniques to drive parser actions. Several other recent parsers use statistics of pairs of head words in conjunction with chart parsing techniques to achieve high accuracy. Collins (1996, 1997) uses chart parsing ....

....in detail, Marcus (1980) uses shift reduce parsing techniques for natural language, and Briscoe and Carroll (1993) describe probabilistic approaches to LR parsing, a type of shift reduce parsing. Other recent machine learning approaches to shift reduce parsing include Magerman (1995) and Hermjakob and Mooney (1997). 3.2. Probability Models that Use Context to Predict Parsing Actions The parser uses a history based approach (Black et al. 1993) in which a probability pX (ajb) is used to score an action a of procedure X 2 f tag, chunk, build, check g, depending on the partial derivation b (also called a ....

[Article contains additional citation context not shown here]

Hermjakob, U., and Mooney, R. J. (1997). Learning Parse and Translation Decision From Examples With Rich Context. In Proceedings of the 35th Annual Meeting of the ACL, and 8th Conference of the EACL Madrid, Spain. ACL.

....treebank for use in a parser, but do not require an explicit hand written grammar. The decision trees in [Black et al. 1993, Jelinek et al. 1994, Magerman, 1995] do not look at words directly, but instead represent words as bitstrings derived from an automatic clustering technique. In contrast, [Hermjakob and Mooney, 1997] uses a rich semantic representation when training decision tree and decision list techniques to drive parser actions. Several other recent papers use statistics of pairs of head words in conjunction with chart parsing techniques to achieve high accuracy. The parsers in [Collins, 1996, Collins, ....

....in [Magerman, 1995] It shows that the maximum entropy parser compares favorably to other state of the art systems [Magerman, 1995, Collins, 1996, Goodman, 1997, Charniak, 1997, Collins, 1997] and shows that only the results of [Collins, 1997] are better in both precision and recall. The parser of [Hermjakob and Mooney, 1997] also performs well (90 labelled precision and recall) on the Wall St. Journal domain, but uses a test set comprised of sentences with only frequent words and recovers a different form of annotation, and is therefore not comparable to the parsers in Table 6.10. Figure 6.11 shows the effects of ....

[Article contains additional citation context not shown here]

Hermjakob, U. and Mooney, R. J. (1997). Learning Parse and Translation Decision From Examples With Rich Context. In Proceedings of the 35th Annual Meeting of the ACL, and 8th Conference of the EACL, Madrid, Spain. ACL.

....that fairly good word for word alignments are recoverable from bilingual text, it remains to be seen whether accurate syntactic alignments are similarly recoverable, and whether those alignments yield reasonable translations. Yet another possibility is to bring a human linguist back into the loop [Hermjakob and Mooney, 1997] as a source of correct parse and transfer decisions. The linguist also supplies general features that are useful for learning to make good decisions in new contexts. 4 Semantics Based Translation Semantics based MT has already produced high quality translations in circumscribed domains. Its ....

Hermjakob, Ulf and Raymond Mooney. 1997. Learning parse and translation decisions from examples with rich context. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics (ACL).

No context found.

Hermjakob, U. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. Ph.D. dissertation, University of Texas at Austin. file://ftp.cs.utexas.edu/pub/ mooney/papers/hermjakobdissertation -97.ps.gz.

....work on automatically learning QA patterns in parse trees (Section 7) On the other hand, the semantic class of the answer (the Qtarget) is used to good effect (Sections 4 and 6) 4. Parsing CONTEX is a deterministic machine learning based grammar learner parser that was originally built for MT (Hermjakob, 1997). For English, parses of unseen sentences measured 87.6 labeled precision and 88.4 labeled recall, trained on 2048 sentences from the Penn Treebank. Over the past few years it has been extended to Japanese and Korean (Hermjakob, 2000) 4.1 Parsing Questions Accuracy is particularly important ....

Hermjakob, U. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. Ph.D.

....in the right range numerically: Q: How many people live in Chile A: nine We use CONTEX, a parser that is trained on a corpus to return both syntactic and semantic information, to help. CONTEX is a deterministic machine learning based grammar learner parser that was originally built for MT (Hermjakob, 1997; Hermjakob and Mooney, 1997) where a smaller version of CONTEX (lexically restricted English) reached a labeled precision rate of 89.8 when trained on 256 sentences. Over the past few years it has been extended over the past years to handle deployment on new languages, including Japanese and ....

Hermjakob, U. and R.J.Mooney. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. In 35th Proceedings of the Conference of the Association for Computational Linguistics (ACL), 482--489. file://ftp.cs.utexas.edu/pub/mooney/papers/contex -acl-97.ps.gz.

....in the right range numerically: Q: How many people live in Chile A: nine We use CONTEX, a parser that is trained on a corpus to return both syntactic and semantic information, to help. CONTEX is a deterministic machine learning based grammar learner parser that was originally built for MT (Hermjakob, 1997; Hermjakob and Mooney, 1997) where a smaller version of CONTEX (lexically restricted English) reached a labeled precision rate of 89.8 when trained on 256 sentences. Over the past few years it has been extended over the past years to handle deployment on new languages, including Japanese and ....

Hermjakob, U. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. Ph.D. dissertation, University of Texas at Austin. file://ftp.cs.utexas.edu/pub/ mooney/papers/hermjakob-dissertation-97.ps.gz.

....1999) 20.6. 4 Syntactic Parsing Perhaps the most well studied problem in computational linguistics is the syntactic analysis of sentences (see Chapters 4 and 12) In addition to statistical methods that have been successfully applied to this task, decision tree induction (Magerman, 1995; Hermjakob Mooney, 1997; Haruno, Shirai, Ooyama, 1999) rule induction (Brill, 1993) and instance based categorization (Cardie, 1993; Argamon, Dagan, Krymolowski, 1998) have also been successfully employed to learn syntactic parsers. One of the rst learning methods applied to parsing the Wall Street Journal (WSJ) ....

Hermjakob, U., & Mooney, R. J. (1997). Learning parse and translation decisions from examples with rich context. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics (ACL-97), pp. 482-489 Madrid, Spain.

....students, writing treebank consistency checking rules (see section 6) making extensions to the tree to parse action sequence module (see section 4.1) and contributing to the background knowledge and feature set. 4 Learning to Parse We base our training on the machine learning based approach of (Hermjakob Mooney, 1997), allowing however unrestricted text and deriving the parse action sequences required for training from a treebank. The basic mechanism for parsing text into a shallow semantic representation is a shift reduce type parser (Marcus, 1980) that breaks parsing into an ordered sequence of small and ....

.... per sentence, we automatically compute the value for every feature in the feature set, add on the parse action as the proper classification of the parse action example, and then feed these examples into a machine learning program, for which we use an extension of decision trees (Quinlan, 1986; Hermjakob Mooney, 1997). We built our parser incrementally. Starting with a small set of syntactic features that are useful across all languages, early training and testing runs reveal machine learning conflict sets and parsing errors that point to additionally required features and possibly also additional background ....

U. Hermjakob and R. J. Mooney. 1997. Learning Parse and Translation Decisions From Examples With Rich Context. In 35th Proceedings of the ACL, pages 482--489.

No context found.

U. Hermjakob. 1997. Learning Parse and Translation Decisions From Examples With Rich Context.

....we had since the group was started in 1992. Finally I would like to give thanks to Brent Adamson, Ulrich Ehrenberger, Ralf Gerlich, Thomas Gohlert, Mike Hoefelein, Martin Kracklauer, Michael Menth, Florian Mertens, Christoph Puntmann and Joe Sullivan for volunteering as translation evaluators. Ulf Hermjakob The University of Texas at Austin May 1997 v Learning Parse and Translation Decisions From Examples With Rich Context Publication No. Ulf Hermjakob, Ph.D. The University of Texas at Austin, 1997 Supervisor: Raymond J. Mooney The parsing of unrestricted text, with its enormous lexical and structural ambiguity, still poses a great ....

Hermjakob, U., & Mooney, R. J. (1997). Learning parse and translation decisions from examples with rich context. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics Madrid, Spain.

No context found.

Hermjakob, U., & Mooney, R. (1997). Learning Parse and Translation Decisions From Examples With Rich Context. Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics (ACL/EACL97) .

No context found.

U. Hermjakob and R. J. Mooney. Learning parse and translation decisions from examples with rich context. In Proceedings of ACL97, pages 482--489, 1997.

No context found.

U. Hermjakob and R.J. Mooney, Learning Parse and Translation Decisions from Examples with Rich Context, In Proceeding of the 35

No context found.

Hermjakob U. (1997) Learning Parse and Translation Decisions From Examples With Rich Context PhD dissertation. University of Texas, Austin.

No context found.

Hermjakob, U. 1997. Learning Parse and Translation Decisions from Examples with Rich Context. Ph.D. dissertation, University of Texas at Austin. file://ftp.cs.utexas.edu/pub/~mooney/papers/hermjakobdissertation -97.ps.gz.

No context found.

Ulf Hermjakob and Raymond J. Mooney. 1997. Learning parse and translation decisions from examples with rich context. In Proceedings of ACL/EACL'97, pages 482--489.

No context found.

Ulf Hermjakob and Raymond J. Mooney. 1997. Learning parse and translation decisions from examples with rich context. In Proc. of ACL'97, pages 482--489, Madrid, Spain. .

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