Báo cáo khoa học: "Memory-Based Learning: Using Similarity for Smoothing" ppt

... Memory-Based Learning: Using Similarity for Smoothing Jakub Zavrel and Walter Daelemans Computational Linguistics Tilburg ... next section, we will argue that a for- mal operationalization of similarity between events, as provided by MBL, can be used for this purpose. In MBL the similarity metric and feature weighting ... of statistical information is an impor- tant iss...

Ngày tải lên: 31/03/2014, 21:20

... chain of rela- tions. This scheme, informally used previously for combining HYPERNYM with other relations, has not been studied for arbitrary pairs of relations. For example, it seems adequate to ... [1] stands for (Winston et al., 1987), [2] for (Cohen and Losielle, 1988) and [3] for (Huhns and Stephens, 1989). lations. The conclusion of an axiom is identified us- ing an algebra...

Ngày tải lên: 23/03/2014, 16:20

... reducing human effort by selective elicitation of partial word alignment using active learning tech- niques. 3 Active Learning for Word Alignment Active learning attempts to optimize performance by ... M t from current iteration t for scoring the links. Re-training and re-tuning an SMT system for each link at a time is computationally infeasi- ble. We therefore perform batch learning...

Ngày tải lên: 30/03/2014, 21:20

... of others, using two sources of information: lexico- graphic resources and distributional similarity. Lexicographic resources are manually-prepared knowledge bases containing semantic information on ... Area Under the Curve (AUC) for recall in the range of 0-0.45 for all algorithms, given background knowledge (knowl- edge consistently improves performance by a few points for all alg...

Ngày tải lên: 30/03/2014, 21:20

... on the classification task for six possible relations be- tween pairs of events: before, immediately-before, included-by, simultaneous, begins and ends. We fo- cus on the before relation because the ... transitive rule: if run BEFORE fall and fall BEFORE injured then run BEFORE injured This increases the number of relations from 37519 to 45619. Perhaps more importantly for our task, of all...

Ngày tải lên: 31/03/2014, 00:20

... (hlh2hs)P(flh3) where ~'~)q(hlh2h3) = 1 for all histories hlhshs. The optimal values for the A~ functions can be estimated using the forward-backward algorithm (Baum, 1972). A decision-tree ... the model. No information about the legal tags for a word are extracted from the test corpus. In fact, no information other than the words is used from the test corpus. For t...

Ngày tải lên: 20/02/2014, 22:20

... decision rule is optimal for the sentence or string error rate. It is not necessarily optimal for other evaluation metrics as for example the BLEU score. One reason for the popularity of the ... values for the scal- ing factors. Because search is performed using the maximum approximation, these absolute values are not needed during the translation process. In con- trast to this,...

Ngày tải lên: 08/03/2014, 03:20

... WebLicht's own data exchange format TCF. 5 The TCF Format The D-SPIN Text Corpus Format TCF (Heid et al, 2010) is used by WebLicht as an internal data exchange format. The TCF format allows the combination ... based data formats were developed beside the TCF format (for example, an encoding for lexi- con based data). In order to avoid any confusion of element names between...

Ngày tải lên: 17/03/2014, 00:20

... decision 4 . T he inference stage exploits the relationships for the propagation of the informa- tion and the BN operates by information reinforce- ment to label a pronoun. We applied all precedent rules ... delim- iter and the pronoun. We considered a third of the corpus for training and the remaining for testing. Our experiment was performed using 20-cross val- idation. Table1 summa...

Ngày tải lên: 17/03/2014, 22:20

... 2009. c 2009 ACL and AFNLP Paraphrase Recognition Using Machine Learning to Combine Similarity Measures Prodromos Malakasiotis Department of Informatics Athens University of Economics and Business Patission ... s 2 . Then, for each s  1 , we compute the nine values f j (s  1 , s 2 ), where f j (1 ≤ j ≤ 9) are the string similarity measures. Finally, we lo- cate the s  1 with the b...

Ngày tải lên: 08/03/2014, 01:20