Báo cáo khoa học: "Pivot Approach for Extracting Paraphrase Patterns from Bilingual Corpora" ppt

... Chinawanghaifeng@rdc.toshiba.com.cnAbstract Paraphrase patterns are useful in paraphrase recognition and generation. In this paper, wepresent a pivot approach for extracting para-phrase patterns from bilingual parallel ... 780–788,Columbus, Ohio, USA, June 2008.c2008 Association for Computational LinguisticsPivot Approach for Extracting Paraphrase Patterns from Bilingual CorporaShiqi Zhao1, Haifeng Wang2, Ting ... of paraphrase patterns from 2M bilingual sentence pairs, the precisionof which exceeds 67%. The evaluation re-sults show that: (1) The pivot approach iseffective in extracting paraphrase patterns, which...

Báo cáo khoa học: "Learning Translations of Named-Entity Phrases from Parallel Corpora" ppt

... (See Section 5 for a reviewof prior work in this area.) In this paper, we de-velop a new approach to learning phrase transla-tions from parallel corpora, and show that it per-forms with very ... assistant for a lexi-cographer that proposes technical terms extracted from a corpus using monolingual methods, and for those approved by the user, proposes possibletranslations from a parallel ... task that we take advantage ofto optimize our performance, to suggest how thealgorithm might perform on more general tasks.Finally we compare our approach and results toprevious work on learning...

Tài liệu Báo cáo khoa học: "Natural Language for Expert Systems: Comparisons with Database Systems" ppt

... gathering information relevant to the problem and making deductions based on that information. In some cases, that information is gathered from a system environment, while in others, the information ... opportunity to provide additional relevant information. Even if this information is not strictly necessary for the problem-solving activity, it provides information about the user's plans and ... speaker's intended meaning from a question. Their work was done within the context of a railroad information system, a type of database. As with the Grieean-based work, their approach is being carried...

Báo cáo khoa học: "Accurate Learning for Chinese Function Tags from Minimal Features" pdf

... simplyextend previous method from English toChinese, we also proposed an effectiveway to recognize function tags directly from lexical information, which is eas-ily scalable for languages that lack ... yT, from a given sequence of input words x =x1, , xT, where yi∈ Σ. Therefore the functionlabeling task can be formulated as a stream of se-quence learning problem. The general approach is ... combinationFT7 gains the best results for all three modelswe considered. As we have expected, the perfor-mance improves as the context window expanded from 2 to 4 (from FT1 to FT3 in Figure 3). Thesentence...

Báo cáo khoa học: "Knowledge acquisition for a constrained speech system using WoZ" pptx

...

Báo cáo khoa học: "Distributional Representations for Handling Sparsity in Supervised Sequence-Labeling" pptx

... trend is for HMM smoothing to outperform LSA, which inturn outperforms TF and TF-IDF. HMM taggingperformance on OOV terms improves by 11%, andchunking performance by 21%. Tagging perfor-mance ... v|xi). For exam-ple, the left context for “reformulated” in our ex-ample above would contain a nonzero probability for the word “of.” Using the features F(xi), a se-quence labeler can learn patterns ... <.001 for all words and for OOV words, using the Chi-squaredtest with 1 degree of freedom.1% and 4%, respectively, for NP chunks in the testset from the original domain. The test set for tag-ging...

Báo cáo khoa học: "AN APPROACH TO SENTENCE-LEVEL TRANSLATION ANAPHORA IN MACHINE" ppt

... of rule that is quite different from the ordinary rules. This distinguishes the system from, for example DCG. With PATR and DCG the possibility of percolation from, say topic to trace, influ- ... Although other approaches, such as direct movement, are feasible too, we win adhere to the comp-to-comp approach. Data from Spanish (Torrego 1984) also seem to support the preference for actual ... to MiMo but makes a uniform account of several phenomena possi- ble. This type of cornplementizer has two slots: an optional slot for topics or wh-words and a slot for a verb construc- tion....

Báo cáo khoa học: "Error Detection for Statistical Machine Translation Using Linguistic Features" pptx

... distinguishing incorrectparts from correct parts is therefore very desir-able not only for post-editing and interactive ma-chine translation (Uefﬁng and Ney, 2007) but also for SMT itself: either ... SMT systems or from system outputs.To some extent, it has already been considered bySMT systems. Hence ﬁnding external information604sources from outside SMT systems is desired for error detection.Linguistic ... n-gramlanguage model for word-level and sentence-levelCE. They also explore linguistic features using in-formation from syntactic category, tense, genderand so on. Unfortunately, such linguistic...

Báo cáo khoa học: "Machine Methods for Proving Logical Arguments Expressed in Englis" ppt

... the form NP + $1/ PRNAME; go to 5. 4.2. No: go to 5. 5. Enter SFORM, and determine quasi-logical form of parsed formula; enter LF, and determine logical translation of quasi-logical formula; ... PHI/.n in the formula on Shelf 12 with copies of the formula in the workspace; go to 6. 6. Read in next PHI from Shelf 9. 6.1. Succeed: go to 5. 6.2. Fail: transfer formula from Shelf 12 ... mark end of formula; go to 2. 7. Combine formulae on Shelf 24 into a single formula of conditional form, in which the conjunction of the premisses implies the conclusion; store formula on Shelf...

Báo cáo khoa học: "Generative Models for Statistical Parsing with Combinatory Categorial Grammar" pptx

... ap-propriate for a formalism like CCG with a large setof lexical categories than one generic token for allunknown words.The performance of the baseline model is shownin the top row of table 3. For ... performance, nor does it alle-viate the problem for sentences with missing lexicalentries for seen words.5We compute λ in the same way as Collins (1999), p. 185.6Smoothing was only done for ... grammars, distin-guishing for example between many classes of verbswith different subcategorization frames. As a re-sult, the categorial lexicon extracted for this purpose from the training corpus...

Xem thêm