UNAL-NLP: Combining Soft Cardinality Features for Semantic Textual Similarity, Relatedness and Entailment

Proceedings of the 8th International Workshop on Semantic Evaluation (SemEval 2014), pages 732–742, Dublin, Ireland, August 23-24, 2014. UNAL-NLP: Combining Soft Cardinality Features for Semantic Textual Similarity, Relatedness and Entailment Sergio Jimenez, George Due˜nas, and Julia Baquero Universidad Nacional de Colombia Ciudad Universitaria, edificio 453, oficina 114, Bogot´a, Colombia [sgjimenezv,geduenasl, jmbaquerov]@unal.edu.co Alexander Gelbukh Center for Computing Research (CIC), Instituto Polit´ecnico Nacional (IPN), Av. Juan Dios B´atiz, Av. Mendiz´abal, Col. Nueva Industrial Vallejo, Mexico City, Mexico www.gelbukh.com Abstract This paper describes our participation in the SemEval-2014 tasks 1, 3 and 10. We used an uniform approach for addressing all the tasks using the soft cardinality for extracting features from text pairs, and machine learning for predicting the gold standards. Our submitted systems ranked among the top systems in all the task and sub-tasks in which we participated. These results confirm the results obtained in pre- vious SemEval campaigns suggesting that the soft cardinality is a simple and useful tool for addressing a wide range of natural language processing problems. 1 Introduction The semantic textual similarity is a core prob- lem in the computational linguistic field. Con- sequently, the previous evaluation campaigns of this task in SemEval have attracted the attention of many research groups worldwide (Agirre et al., 2012; Agirre et al., 2013).This year, 3 tasks related to this problem have been proposed exploring dif- ferent facets such as semantic relatedness, entail- ment , multilingualism, lack of training data and imbalance in the amount of information. The soft cardinality (Jimenez et al., 2010) is a simple concept that generalizes the classical set cardinality by considering the similarities among the elements in a collection for a more intuitive quantification of the number of elements in that collection. This approach can be applied to text applications representing texts as collections of words and providing a similarity function that compares two words. Varying this word-to-word similarity function the soft cardinality can reflect This work is licensed under a Creative Commons At- tribution 4.0 International Licence. Page numbers and pro- ceedings footer are added by the organisers. Licence details: http://creativecommons.org/licenses/by/4.0/ notions of syntactic similarity, semantic related- ness, among others. We (and others) have used this approach to address with success the semantic textual similarity and other tasks in previous Se- mEval editions (Jimenez et al., 2012b; Jimenez et al., 2012a; Jimenez et al., 2013a; Jimenez et al., 2013b; Jimenez et al., 2013c; Croce et al., 2013). In this paper we describe our participating sys- tems in the SemEval-2014 tasks 1, 3, and 10, which used the soft cardinality as core approach. 2 Features from Soft Cardinalities The cardinality of a collection of elements is the counting of non-repeated elements in it. This def- inition is intrinsically associated with the notion of set, which is a collection of non-repeated ele- ments.Thus, the cardinality of a collection or set A is denoted as |A|. Clearly, the cardinality of a collection with repeated elements treats groups of identical elements as a single instance contribut- ing only with a …

Content truncated to 500 words