Adaptation of a Machine-learning Textual Entailment System to a Multilingual Answer Validation Exercise

Zornitsa Kozareva, Sonia Vázquez, Andrés Montoyo
2006 Conference and Labs of the Evaluation Forum  
The recognition of textual entailment is a new generic task in which a system automatically establishes whether a given text entails the meaning of another. This detection is useful for many NLP applications. In the case of Question Answering, textual entailment is represented as an answer validation exercise (AVE) where the system has to verify the correctness of the returned snippet of the Question Answering system. In this paper we present an AVE system based on the combination of word
more » ... p and Latent Semantic Indexing modules. The main focus and contribution of our work consist in the adaptation of the already developed and evaluated textual entailment system MLEnt and its ability to function with different languages. We have evaluated our approach with the English, Spanish, French, Dutch, German, Italian and Portuguese languages.
dblp:conf/clef/KozarevaVM06a fatcat:j5xpjjgrardmtcyzqinndcpd2m