a mentionsynchronous coreference resolution algorithm

Ngày tải lên: 31/03/2014, 03:20

Ngày tải lên: 31/03/2014, 20:20

... (which can be easily incorporated into the coreference resolution process as features), different resolution algorithms that aim at enhancing global coherence of coreference chains, and also adapting BART ... adapting BART to different corpora. Through the availability of BART as open source, as well as its modularity and adaptability, we hope to create a larger com- munity that allows both to push the state ... for standoff annotation allows the use of the coreference resolution as part of a larger sys- tem, but also performing qualitative error analysis using integrated MMAX2 functionality (annotation 1 An...

Ngày tải lên: 20/02/2014, 09:20

... non- parametric problems. Annals of Statistics. Niyu Ge, John Hale, and Eugene Charniak. 1998. A statistical approach to anaphora resolution. Sixth Workshop on Very Large Corpora. Xiaoqiang Luo, Abe ... JMLR. Pascal Denis and Jason Baldridge. 2007. Global, joint determi- nation of anaphoricity and coreference resolution using inte- ger programming. HLT-NAACL. Thomas Ferguson. 1973. A bayesian analysis ... salience feature. This model fixes many anaphora errors and in particular fixes the second anaphora error in figure 2(c). 3.5 Cross Document Coreference One advantage of a fully generative approach...

Ngày tải lên: 08/03/2014, 02:21

Ngày tải lên: 22/06/2014, 23:20

... Related Work Several coreference resolution systems are cur- rently publicly available. JavaRap (Qiu et al., 2004) is an implementation of the Lappin and Leass’ (1994) Resolution of Anaphora Procedure (RAP). ... reference implementation of the rap anaphora resolution algorithm. In Proceedings of the Language Resources and Evaluation Conference. W. Soon, H. Ng, and D. Lim. 2001. A Machine Learning Ap- proach to Coreference ... state-of-the-art. Coreference resolution has received much re- search attention, resulting in an array of ap- proaches, algorithms and features. Reconcile is modeled after typical supervised learning ap- proaches...

Ngày tải lên: 20/02/2014, 04:20

... the ACL Student Research Workshop, pages 19–24, Ann Arbor, Michigan. Zoubin Gha hramani and Michael I. Jordan. 1997. Facto- rial hidden markov models. Machine Learning, 29:1– 31. A. Haghighi and ... Ohio schuler@ling.osu.edu Abstract This paper presents a supervised pronoun anaphora resolution system based on factorial hidden Markov models (FHMMs). The ba- sic idea is that the hidden states of FHMMs are a n explicit ... morphological features of words trained from the corpus and the strings concatenated from the tree leaves are made. This method is about as accurate as the approach described by Klein and Manning...

Ngày tải lên: 20/02/2014, 04:20

... rates. By default, the ALEPH algorithm only generates rules that have 100% accuracy for the training data. And each rule contains at most three predicates. To accommodate for coreference resolution, ... restrictions to allow rules that have above 50% accuracy and contain up to ten predicates. De- fault parameters were applied for all the other set- tings in ALEPH as well as other learning algorithms used ... automated and manual acquisition of anaphora resolution strate- gies. In Proceedings of the 33rd Annual Meeting of the Association for Computational Linguistics (ACL), pages 122–129. V. Claveau, P....

Ngày tải lên: 20/02/2014, 09:20

... ideal, all such available systems use verbal features that are inap- plicable to our dataset, such as punctuation, capital- ization, and gazetteers. The verbal features that we have included are ... finite-state transducer. Acknowledgments We thank Aaron Adler, Regina Barzilay, S. R. K. Branavan, Sonya Cates, Erdong Chen, Michael Collins, Lisa Guttentag, Michael Oltmans, and Tom Ouyang. This ... Because we are working from transcripts rather than text, features dependent on punctuation and capitalization, such as apposition, are also not applicable. 4.3 Non-verbal features Our non-verbal...

Ngày tải lên: 20/02/2014, 12:20

... Rank Candidate Partitions We train an SVM-based ranker for ranking candidate partitions by means of Joachims’ (2002) SVM package, with all the parameters set to their default values. To create ... HLT/NAACL, pages 297–304. A. Berger, S. Della Pietra, and V. Della Pietra. 1996. A maximum entropy approach to natural language pro- cessing. Computational Linguistics, 22(1):39–71. C. Cardie and ... reviewers for their valuable comments on an earlier draft of the paper. References C. Aone and S. W. Bennett. 1995. Evaluating automated and manual acquisition of anaphora resolution strate- gies. In...

Ngày tải lên: 20/02/2014, 15:20

... reduce this search space, but we also avoid strong commitments. 3 The Algorithm at a Glance Basically, the best-first search algorithm is a generalization of the incremental version: instead of successively ... 2. Descriptors are organized in a taxonomy, to capture generalizations; the associated redun- dancies are exploited in the selection process. 3. Descriptor combinations are limited in size. 4. Negations are ... unused at each state. The expansion process is guided by linguistically motivated preferences (1 and 2, from (Dale, Reiter 1995), adapted to boolean combinations): 1. First, a boolean combination...

Ngày tải lên: 22/02/2014, 02:20

... of Texas at Dallas Richardson, TX 75083, USA Abstract This paper examines how a new class of nonparametric Bayesian models can be ef- fectively applied to an open-domain event coreference task. ... the Fourth Intern ational Workshop on Semantic Evalu- ations (SemEva l), pages 460–4 63. Cosmin Adrian Bejan, Matthew Titsworth, Andrew Hickl, and San da Harabagiu. 2009. Nonparametric Bayesian Models ... param- eters associated with an event z, φ a notation for all model parameters, and X a notation for all ran- dom variables that represent observable features. 2 Given a document collection annotated...

Ngày tải lên: 07/03/2014, 22:20

... that the features that work best for each language largely overlap and are language inde- pendent, like head match, is -a match, and whether the mentions are pronominal. 5 Parameter 2: Annotation ... S. Lalitha Devi, A. Branco, and R. Mitkov, editors, Anaphora Processing and Ap- plications (DAARC 2009), volume 5847 of LNAI, pages 29–42. Springer-Verlag. Marta Recasens and M. Ant ` onia Mart ´ ı. ... 2009. AnCora- CO: Coreferentially annotated corpora for Spanish and Catalan. Language Resources and Evaluation, DOI 10.1007/s10579-009-9108-x. Marta Recasens, Llu ´ ıs M ` arquez, Emili Sapena, M....

Ngày tải lên: 07/03/2014, 22:20

... Quinlan. 1993. C4.5: Programs for machine learning. Morgan Kaufmann Publishers Inc., San Francisco, CA, USA. Karthik Raghunathan, Heeyoung Lee, Sudarshan Ran- garajan, Nathanael Chambers, Mihai ... Kessler, and Gunther Heidemann. 2010. Re- lational feature engineering of natural language pro- cessing. In CIKM ’10. ACM. Xiaoqiang Luo, Abe Ittycheriah, Hongyan Jing, Nanda Kambhatla, and Salim ... Release 2.0 (Pradhan et al., 2007). The ad- vantage of OntoNotes is that it contains two parallel annotations: (i) a gold setting, gold standard manual annotations of the preprocessing information...

Ngày tải lên: 07/03/2014, 22:20

... resolu- tion, namely, large-scale knowledge bases (YAGO, FrameNet), coreference- annotated data (Noun Pairs, Verb Pairs), and unannotated data (Appositives), by applying them to two learning-based coreference models, ... addition to coreference- annotated data and unannotated data; (2) incorporate such knowledge as features into a richer baseline feature set that we previously employed (Rahman and Ng, 2009); and ... (coreference- annotated) train- ing data and use them as features to train a resolver. However, for these features to be effective, we need to address data sparseness, as many noun pairs in the training data may...

Ngày tải lên: 07/03/2014, 22:20

... w 1 w 2 3 single-character word w 4 a word starting with character c and having length l 5 a word ending with character c and having length l 6 space-separated characters c 1 and c 2 7 character bigram c 1 c 2 in ... the source agenda and the target agenda. Initially the source agenda contains an empty sentence and the target agenda is empty. At each processing stage, the decoder reads in a character from ... with each candidate in the source agenda and puts the generated candidates onto the target agenda. After each character is processed, the items in the target agenda are copied to the source agenda,...

Ngày tải lên: 08/03/2014, 02:21

... in the training data of the MERGED dataset. SRL features show a high χ 2 value, ranking immediately after string matching and alias, which indicates a high corre- lation of these features to ... leaving out each feature in turn. Again, it can be seen that remov- ing both I and J SEMROLE induces a relatively high performance degradation when compared to other features. Their removal ranks ... Germany http://www.eml-research.de/nlp/ Abstract Extending a machine learning based coref- erence resolution system with a feature capturing automatically generated infor- mation about semantic roles improves its performance. 1 Introduction The last...

Ngày tải lên: 08/03/2014, 21:20

... more efficiently than do alternating ATAT tracts [6,7], resulting in a preference for binding to the DNA in the order AATT > TAAT ¼ TTAA ¼ ATAT > TATA [8]. For AATT tracts, binding of the drug can be ... N (2000) Local conformational variations observed in B-DNA crystals do not improve base stacking. Computational analysis of base stacking in d(CATGGGCCCATG) 2 B M A intermediate crystal structure. ... d(CGCGTTAACGCG) with an inverted central tetranucleotide and its netropsin complex. Acta Cryst D51 , 190–198. 17 Abrescia NGA, Malinina L & Subirana JA (1999) Stacking interaction of guanine...

Ngày tải lên: 16/03/2014, 22:20

Ngày tải lên: 17/03/2014, 02:20

Ngày tải lên: 17/03/2014, 04:20