Báo cáo khoa học: "Weakly Supervised Part-of-Speech Tagging for Morphologically-Rich, Resource-Scarce Languages" potx

... Greece, 30 March – 3 April 2009. c 2009 Association for Computational Linguistics Weakly Supervised Part-of-Speech Tagging for Morphologically-Rich, Resource-Scarce Languages Kazi Saidul Hasan and ... 75083-0688 {saidul,vince}@hlt.utdallas.edu Abstract This paper examines unsupervised ap- proaches to part-of-speech (POS) tagging for morphologically-rich, resource-scarc...

Ngày tải lên: 24/03/2014, 03:20

... time distributions for information extraction, in general, and NE extrac- tion, in particular, is not new. (Hetland, 2004) surveys recent methods for scoring time sequences for similarity. (Shinyama ... accuracy drop due to misalignment could be accommodated for by using a larger temporal 822 Figure 5: Transliteration lists for forsyth for two iterations of the algorithm. As trans...

Ngày tải lên: 20/02/2014, 12:20

... basis for finding and further discriminating more detailed semantic relations. 3 A Weakly Supervised Approach to Learning Presupposition Relations We describe a weakly supervised approach for learning ... in context. For this purpose we randomly selected 10 instances for each verb pair type (for rare verb pair types only 5). We compared the gold standards elaborated by the same ju...

Ngày tải lên: 07/03/2014, 22:20

... (Davidov et al., 2007) are only evaluated for three hand-picked classes, with preci- sion scores of 90% for names of countries, 87% for fish species and 68% for instances of constellations. Our evaluation ... queries based on vectors {V} 04. For each cluster of phrases S in {S} 05. {C S } = ∅ 06. For each query Q of S 07. Insert labels of Q from {M E } into {C S } 08. For each la...

Ngày tải lên: 08/03/2014, 01:20

... of 53% for E s and 304% for E a . 4.2 Evaluation of Attributes Target Hierarchy Concepts: The performance of attribute extraction is assessed over a set of 25 tar- get concepts also used for evaluation ... Seed attributes are provided as input for only one target concept (top graphs), or for each target concept (bottom graphs) for failing to extract any attributes for some tar-...

Ngày tải lên: 08/03/2014, 21:20

... hybrid co/self-training strategy for weakly supervised re- lation classification and (Chen et al., 2006) where a graph based algorithm called label propagation is employed to perform weakly supervised relation ... Model for Training Data Acquisition In this section, we derive a simple probabilistic model for acquiring training data for a given learn- ing task, and use it to motivat...

Ngày tải lên: 23/03/2014, 18:20

... Weakly Supervised Learning for Cross-document Person Name Disambiguation Supported by Information Extraction Cheng Niu, Wei Li, and Rohini K. ... name disambiguation calls for NLP/IE support which may have extracted the key person’s identification information from the textual documents. compared. Significant performance enhancement up ... probabilities in Eq. (3), for a given...

Ngày tải lên: 31/03/2014, 03:20

... the results for the same approach but us- ing second-order features for the person category. The Class-Pattern approach showed low perfor- mance, similar to the random classification, for which ... them. This makes our weakly supervised methodology appli- cable on larger scale than supervised approaches, still having significantly better performance than the unsupervised ones. In our...

Ngày tải lên: 31/03/2014, 20:20

... the Association for Computational Linguistics:shortpapers, pages 42–47, Portland, Oregon, June 19-24, 2011. c 2011 Association for Computational Linguistics Part-of-Speech Tagging for Twitter: ... have been shown to perform poorly on Twitter (Finin et al., 2010). One of the most fundamental parts of the linguis- tic pipeline is part-of-speech (POS) tagging, a basic form of synta...

Ngày tải lên: 20/02/2014, 04:20

... we mentioned before. Note that in (10) we have split the regularizer into two parts; one for the supervised component of the objective, and the other for the unsupervised component. Thus the semi -supervised convex ... from semi -supervised data that significantly outper- forms corresponding supervised methods. 1 Introduction Supervised learning algorithms still represent the state...

Ngày tải lên: 08/03/2014, 01:20