TY - GEN
T1 - Recognizing textual entailment with statistical methods
AU - Ríos Gaona, Miguel Angel
AU - Gelbukh, Alexander
AU - Bandyopadhyay, Sivaji
PY - 2010
Y1 - 2010
N2 - In this paper we propose a new cause-effect non-symmetric measure applied to the task of Recognizing Textual Entailment .First we searched over a big corpus for sentences which contains the discourse marker "because" and collected cause-effect pairs. The entailment recognition is based on measure the cause-effect relation between the text and the hypothesis using the relative frequencies of words from the cause-effect pairs. Our measure outperformed the baseline method, over the three test sets of the PASCAL Recognizing Textual Entailment Challenges (RTE). The measure shows to be good at discriminate over the "true" class. Therefore we develop a meta-classifier using a symmetric measure and a non-symmetric measure as base classifiers. So, our meta-classifier has a competitive performance.
AB - In this paper we propose a new cause-effect non-symmetric measure applied to the task of Recognizing Textual Entailment .First we searched over a big corpus for sentences which contains the discourse marker "because" and collected cause-effect pairs. The entailment recognition is based on measure the cause-effect relation between the text and the hypothesis using the relative frequencies of words from the cause-effect pairs. Our measure outperformed the baseline method, over the three test sets of the PASCAL Recognizing Textual Entailment Challenges (RTE). The measure shows to be good at discriminate over the "true" class. Therefore we develop a meta-classifier using a symmetric measure and a non-symmetric measure as base classifiers. So, our meta-classifier has a competitive performance.
UR - http://www.scopus.com/inward/record.url?scp=78751474244&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-15992-3_39
DO - 10.1007/978-3-642-15992-3_39
M3 - Contribución a la conferencia
SN - 3642159915
SN - 9783642159916
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 372
EP - 381
BT - Advances in Pattern Recognition - Second Mexican Conference on Pattern Recognition, MCPR 2010, Proceedings
T2 - Mexican Conference on Pattern Recognition 2010, MCPR 2010
Y2 - 27 September 2010 through 29 September 2010
ER -