Báo cáo khoa học: "Semi-supervised Convex Training for Dependency Parsing" potx

... allows that use of future information in ad- dition to past information. The pipeline model nat- urally allows access to all the past information. Algorithm 3 Pseudo code for the look ahead algo- rithm. ... develop a framework for decisions made via in pipeline models, which addresses these difficulties, and presents and evaluates it in the context of bottom up dependency parsing for E...

Ngày tải lên: 20/02/2014, 12:20

... structures in dependency pars- ing are dependency graphs. Therefore, as items for constituency parsers are defined as sets of par- tial constituency trees, it is tempting to define items for dependency ... i.e., the set of forests containing a single dependency tree of the form w i (w i ). This same set of hypotheses can be used for all the parsers, so we will not make it ex- plic...

Ngày tải lên: 08/03/2014, 01:20

... algorithm for structured SVMs which is convex optimization. Fi- nally, we apply this algorithm to dependency pars- ing and show improved dependency parsing accu- racy for both Chinese and English. 2 Dependency ... is non -convex. Therefore the objective as a whole is non -convex, making the search for global optimal difficult. Note that the root of the optimiza- tion difficulty f...

Ngày tải lên: 08/03/2014, 01:20

... the Association for Computational Linguistics:shortpapers, pages 323–328, Portland, Oregon, June 19-24, 2011. c 2011 Association for Computational Linguistics Models and Training for Unsupervised ... HLT 2010 First International Workshop on Formalisms and Methodology for Learning by Reading. Association for Computational Linguistics, Los Angeles, Califor- nia, June. Tom O’Hara an...

Ngày tải lên: 20/02/2014, 05:20

... In addition to performance, efficiency of the classifier (C) is important for self -training, which is computationally expensive. Our classifier is a compromise between performance and efficien- cy. ... set. We use one unlabeled training set (U) for in- domain and another for out-of-domain experi- ments. The former is the remaining portion of CoNLL training data and contains 3...

Ngày tải lên: 07/03/2014, 22:20

... 1 Enlarged Training Data for Language 2 Manually Prepared Training Data for Language 2 ClassifierClassifier Further Enlarged Training Data for Language 1 Further Enlarged Training Data for Language ... of co -training (Blum and Mitchell, 1998). Language 1 Language 2 Iteration Manually Prepared Training Data for Language 1 Classifier Classifier Training Training Enlarged...

Ngày tải lên: 08/03/2014, 00:20

... phrase translation is formulated as an information re- trieval process implemented with a log-linear model aiming for a balanced precision and re- call. We present a generic phrase training al- gorithm ... overestimation problem in HMM especially when training data size is small. 3.2 Bilingual Information Metric Trying to find phrase translations for any possible n- gram is not a good...

Ngày tải lên: 08/03/2014, 01:20

... a form of gap-threading, but the structures generated by the grammar are strictly projective. Moreover, the study of formal gram- mars is only partially relevant for research on data- driven dependency ... the com- putation of model predictions or scores for such arcs. The parsing experiments are performed with a variant of Covington’s algorithm for dependency parsing (Covington, 20...

Ngày tải lên: 22/02/2014, 02:20

... argued for a dependency formal- ism called Discontinuous Grammar and annotated a large set of Danish sentences using this formal- ism to create the Danish Dependency Treebank (Kromann, 2003). The formalism ... to state-of-the-art accuracy for English and the best accuracy we know of for Czech and Danish. 2 Maximum Spanning Tree Parsing Dependency- tree parsing as the search for t...

Ngày tải lên: 22/02/2014, 02:20

... conversion process, consider the sentence “A big crowd turned out for the parade.” The dependency graph for this, as generated by the Stanford dependency parser, is shown in Figure 3. The following relations ... root dependency. The dependent is added to the diagram data structure based on its relation to the governor. Before analysis of dependency graphs begins, our algorithm take...

Ngày tải lên: 08/03/2014, 01:20