... error-driven pruning formachinelearning of coreference rules. In Proc. of EMNLP, pages 55–62.V. Ng and C. Cardie. 2002b. Improving machine learn-ing approaches to coreference resolution. In Proc. ... 104–111.J. R. Quinlan. 1993. C4.5: Programs for Machine Learning. Morgan Kaufmann.W. M. Soon, H. T. Ng, and D. Lim. 2001. A machine learning approach to coreferenceresolution of nounphrases. Computational ... 2003. Coref-erence resolutionusing competitive learningapproach.In Proc. of the ACL, pages 176–183.D. Zelenko, C. Aone, and J. Tibbetts. 2004. Coreference resolutionfor information extraction....
... to coreference resolution. In Proceed-ings of the 40th Annual Meeting of the Association for Computational Linguistics (ACL), pages 104–111,Philadelphia.V. Ng. 2005. Machinelearningforcoreference ... 65.4 63.5Table 4: Results of different systems forcoreference resolution examined the C4.5 algorithm4which is widely used for the coreferenceresolution task. The first line ofthe table shows ... Powell”, and therefore refers to a maleperson and cannot co-refer with “she”.The entity-mention model based on Eq. (2) per-forms coreferenceresolution at an entity-level. For simplicity, the...
... application-internal representationsto a suitable format for several machine learning toolkits: One module exposes the functionality ofthe the WEKA machinelearning toolkit (Wittenand Frank, 2005), ... using it as a platform for research including the use of new informationsources (which can be easily incorporated into the coreference resolution process as features), different resolution algorithms ... decision trees for coreference resolution. In Proc. IJCAI 1995.Morton, T. S. (2000). Coreferencefor NLP applications. InProc. ACL 2000.Moschitti, A. (2006). Making tree kernels practical for naturallanguage...
... en-tity. Coreferenceresolution on text datasets is well-studied (e.g., (Cardie and Wagstaff, 1999)). Thisprior work provides the departure point for our in-vestigation of coreferenceresolution ... documents.Evaluation metric Coreferenceresolution is of-ten performed in two phases: a binary classifi-cation phase, in which the likelihood of corefer-ence for each pair of noun phrases ... 352–359,Prague, Czech Republic, June 2007.c2007 Association for Computational LinguisticsConditional Modality Fusion forCoreference Resolution Jacob Eisenstein and Randall DavisComputer Science...
... expressions is given in Table 1.2.2 Learning Algorithm For learningcoreference decisions, we used aMaximum Entropy (Berger et al., 1996) model. Coreference resolution is viewed as a binary clas-sification ... of machinelearning based coreference resolution systems (Soon et al., 2001;Ng & Cardie, 2002; Kehler et al., 2004, inter alia).Similarly, many researchers have explored tech-niques for ... Germanyhttp://www.eml-research.de/nlp/AbstractExtending a machinelearning based coref-erence resolution system with a featurecapturing automatically generated infor-mation about semantic roles improves itsperformance.1 IntroductionThe...
... used to build a machine learning process. The notion of observing data, learning from it, and thenautomating some process of recognition is at the heart of machinelearning and formsthe primary ... exploring machinelearning withR! Before we proceed to the case studies, however, we will review some R functionsand operations that we will use frequently.R Basics forMachine Learning As ... message that is printed when you draw theR forMachineLearning | 19www.it-ebooks.infoWith the function defined, we will use the lapply function, short for “list-apply,” toiterate this function...
... increases in F-measure for both classi-fiers and both data sets. When using RIPPER, for example, performance increases from 64.3 to 67.2 for the MUC-6 data set and from 60.8 to 63.2 for MUC-7. Similar, ... Unfortunately, the hand-selectedfeatures precipitate a large drop in precision for pro-noun resolutionfor the MUC-7/C4.5 data set. Ad-ditional analysis is required to determine the reason for ... improve precision on common noun resolution. Overall, the learning framework and lin-guistic knowledge source modifications boost per-formance of Soon’s learning- based coreference res-olution approach...
... Association for Computational Linguistics.Simon Tong and Daphne Koller. 2002. Support vec-tor machine active learning with applications to textclassification. Journal of MachineLearning Re-search ... Number of Foreign Words AnnotatedBLEU ScoreNumber of Foreign Words Annotatedthe approx. 54,500 foreign wordswe selectively sampled for annotation cost = $205.80last approx. 700,000 foreign ... preference for covering frequent n-grams before covering in-frequent n-grams. The VG method is depicted inFigure 2.Figure 3 shows the learning curves for bothjHier and jSyntax for VG selection...
... correspond.To solve the former problem, we apply a maxi-mum entropy model to Knight and Marcu’s modelto introduce machinelearning features that are de-fined not only for CFG rules but also for othercharacteristics ... 850–857,Sydney, July 2006.c2006 Association for Computational LinguisticsTrimming CFG Parse Trees for Sentence Compression Using Machine Learning ApproachesYuya Unno1Takashi Ninomiya2Yusuke ... cre-ate a compression forest as Knight and Marcu did.We select the tree assigned the highest probabilityfrom the forest.Features in the maximum entropy model are de-fined for a tree node and...
... (Daelemans et al., 2004) for Memory-Based Learning, the MaxEnt Toolkit (Le, 2004) for Maximum Entropy and LIBSVM (Chang andLin, 2001) for Support Vector Machines. For TiMBL we used k nearest ... performance for gold-standard treesscoring 89.34% on accuracy and 86.87% on f-score. The learning curves for the three algo-rithms, shown in Figure 4, are also informative,with SVM outperforming ... memory-based learning toperform various graph transformations. One of thetransformations is node relabelling, which addsfunction tags to parser output. They report an f-score of 88.5% for the...
... 25–32,Prague, Czech Republic, June 2007.c2007 Association for Computational LinguisticsTransductive learningfor statistical machine translationNicola UeffingNational Research Council CanadaGatineau, ... relevant for translating the new data arereinforced. The probability distribution over thephrase pairs thus gets more focused on the (reliable)parts which are relevant for the test data. For an ... Semi-supervised training for statistical word alignment. In Proc. ACL.S. Nießen, F. J. Och, G. Leusch, and H. Ney. 2000. Anevaluation tool formachine translation: Fast evalua-tion for MT research....
... Pittsburgh{jsa8,hwa}@cs.pitt.eduAbstract Recent studies suggest that machine learn-ing can be applied to develop good auto-matic evaluation metrics formachine trans-lated sentences. This paper further ana-lyzes aspects of learning ... criteria. Machinelearning af-fords a unified framework to compose these crite-ria into a single metric. In this paper, we havedemonstrated the viability of a regression approachto learning ... and Chris Brockett.2001. A machinelearning approach to the automatic eval-uation of machine translation. In Proceedings of the 39thAnnual Meeting of the Association for Computational Lin-guistics,...
... with machine learning algorithms that perform classification, clustering and pattern inductiontasks.• Having a good annotation scheme and accurate annotations are critical for machine learning ... that this is where you start for designing the features that go into your learning algorithm. The better the features, thebetter the performance of the machinelearning algorithm!Preparing ... particular problem or phenomenon that has sparkedyour interest, for which you will need to label natural language data for training for machine learning. Consider two kinds of problems. First imagine...
... 2005.c2005 Association for Computational LinguisticsUsing Emoticons to reduce Dependency in Machine Learning Techniques for Sentiment ClassificationJonathon ReadDepartment of InformaticsUniversity ... language-style dependency.Also, note that neither machine- learning modelconsistently out-performs the other. We speculatethat this, and the generally mediocre performance ofthe classifiers, is due (at ... best-performing settings for the Na¨ıve Bayes classifierwas a window context of 130 tokens taken from thelargest training set of 22,000 articles. Similarly, thebest performance for the SVM...
... T¨uba-D/Z will show whether the performance achieved for the HTC test set is scalable. For future versions of the system, it might also resolution (no values were given for pronouns).Although a number ... learner is thus forced toconcentrate on difficult examples.Although boosting has not yet been appliedto coreference resolution, it has outperformedstateof-the-art systems for NLP tasks such ... described in this pa-per.(McCarthy and Lehnert, 1995) were amongthe first to use machinelearningfor coreference resolution. RESOLVE was trained on data fromMUC-5 English Joint Venture (EJV)...