a machine learning approach to pronoun resolution in spoken dialogue

Ngày tải lên: 23/03/2014, 19:20

... Feature Vector The features used by our system are summarised in Table 4.3. The individual features for anaphor A Machine Learning Approach to German Pronoun Resolution Beata Kouchnir Department ... rule-based and the machine learning approach. For the latter, good results have been achieved with large feature sets (including syntactic, se- mantic, grammatical and morphological informa- tion) ... the an- tecedent dist distance in markables between anaphor and an- tecedent (1 20) same agr same agreement of anaphor and antecedent? same gramrole same grammatical role of anaphor and antecedent? same...

Ngày tải lên: 17/03/2014, 06:20

Ngày tải lên: 31/03/2014, 20:20

Ngày tải lên: 24/03/2014, 03:20

Ngày tải lên: 31/03/2014, 04:20

... information available in the training corpus and modelling the task applying an approach similar to the one that has been applied to tasks like semantic role labeling (Morante et al., 2008) or negation scope ... has been argued that memory-based learning is at an advantage in solving these highly disjunctive learn- ing problems compared to more eager learning that abstract from the examples, as the latter ... harmful in language learning. Machine Learning, Special issue on Natural Language Learning, 34:11–41. W. Daelemans, J. Zavrel, K. Van der Sloot, and A. Van den Bosch. 2007. TiMBL: Tilburg memory based learner,...

Ngày tải lên: 23/03/2014, 13:20

... evaluations seems more appropriate. Such an approach allows taking into account in a flexible and natural way the variety of knowledge sources and processing activities that are involved in ... less relevant parts in order to obtain the representation of an appropriate summary to be eventually translated into natural language. These last phases (i.e., pruning and generation) are, for ... standpoint, and we propose a procedural, rule-based approach to importance evaluation. This novel approach is supported by a prototype experimental system, called importance evaluator, that can...

Ngày tải lên: 09/03/2014, 01:20

... constraints as broadly and un- conditionally applicable hard constraints. Because sources of linguistic information in a learning- based system are represented as features, we can, in con- trast, incorporate ... mentioned in a doc- ument. Machine learning approaches to this prob- lem have been reasonably successful, operating pri- marily by recasting the problem as a classification task (e.g. Aone and Bennett ... learning algorithms investigated: AGREEMENT, ANIMACY, BINDING, and MAXIMALNP. Discussion. In an attempt to gain additional in- sight into the difference in performance between our system and the original Soon...

Ngày tải lên: 17/03/2014, 08:20

Ngày tải lên: 23/03/2014, 17:20

Ngày tải lên: 30/03/2014, 17:20

Ngày tải lên: 31/03/2014, 17:20

Ngày tải lên: 01/04/2014, 00:20

Ngày tải lên: 21/06/2014, 08:20

Ngày tải lên: 09/08/2014, 20:21

... (Cohn and La- pata, 2008) expands the operation set by including insertion, substitution and reordering, and incorpo- rates grammar rules. In speech domain, (Clarke and Lapata, 2008) investigates ... Computational Linguistics A Two-step Approach to Sentence Compression of Spoken Utterances Dong Wang, Xian Qian, Yang Liu The University of Texas at Dallas dongwang,qx,yangl@hlt.utdallas.edu Abstract This ... compression in broadcast news using an integer linear programming approach. There is only a few existing work in spon- taneous speech domains. (Liu and Liu, 2010) mod- eled it as a sequence labeling...

Ngày tải lên: 07/03/2014, 18:20

... objectives and operating model ã An incontestable fact base and cost baseline ã Ongoing transparency and reporting of cost and benefits linked to financial plans ã Clear alignment with existing investment ... Strategic operating model ã Platform replacement / consolidation Reinvesting tactical benefits into transformational initiatives Transformational Tactical = Figure 2: Traditional vs. transformational ... North American banks, on the other hand, are still targeting inefficiencies in the back office. Segment Management Brand Management Channel Integration and Management Marketing, Sales and Servicing Customer management Customer pricing Product aggregation Product pricing Third...

Ngày tải lên: 16/02/2014, 11:20

... syntax tree accord- ing to their relative position in the target language. The ranking model is automatically trained to min- imize the mis-ordering of tree nodes in the training data. Large scale ... English-Japanese machine translation tasks, and experimental results show that our approach can bring significant improvements to the baseline phrase-based SMT system in both pre- ordering and integrated ... Lin. 2008. LIBLINEAR: A library for large lin- ear classification. In Journal of Machine Learning Re- search. Michel Galley, Jonathan Graehl, Kevin Knight, Daniel Marcu, Steve DeNeefe, Wei Wang,...

Ngày tải lên: 19/02/2014, 19:20

... an IAVA can be converted into a query that finds machines in the model that are not in compliance with it. This capability 6 Scaling to large networks can be addressed by adding computing ... CycSecure has proven to be a successful application of AI planning technology in a real-world domain. Logical next steps include applying CycSecure to larger networks and increasing the breadth and ... The Attack Planner enables the user to state plan goals, launch the planner and view the attack plans generated. For example, a user can state the goal “An external user with no initial access...

Ngày tải lên: 05/03/2014, 23:20

... Bioengineering, Indian Institute of Technology Bombay, Powai, Mumbai, India Cellular regulation is a result of complex interactions arising from DNA–protein and protein–protein binding, autoreg- ulation, and ... mechanisms include the tryptophan and arabinose systems of Escherichia coli and the phage lambda switch in bacteriophage. The GAL regulatory network in Saccharomyces cerevis- iae is a prototypical ... regulatory proteins Gal4p, Gal80p, Gal3p and activated Gal3p, respec tively. Model I includes sequen tial bin ding of activated Gal3p to DNA-Gal4p–Gal80p. Parameter m represents cooperativity during...

Ngày tải lên: 07/03/2014, 16:20

... possible facets. Another key characteristic of rich domains like financial analysis, is that facts and events are subject to interpretation in context. To a finan- cial analyst, it makes a difference ... were annotated with the types of financial transactions they are most related to. Paragraphs that did not fall into a category of interest were classified as “other”. The annotated paragraphs ... Maytag: A multi-staged approach to identifying complex events in textual data Conrad Chang, Lisa Ferro, John Gibson, Janet Hitzeman, Suzi Lubar, Justin Palmer, Sean Munson, Marc Vilain, and Benjamin...

Ngày tải lên: 08/03/2014, 21:20