Computer Sciences

Question Classification is an important task in Question Answering Systems. This paper presents a Spanish Question Classi-fier based on machine learning, automatic online translators and different language features. Our system works with Eng-lish collections and bilingual questions (English/Spanish). We have tested two Spanish-English online translators to iden-tify the lost of precision. We have made experiments using lexical, syntactic and semantic features to test which ones made a better performance. The obtained results show that our system makes good classifi-cations, over a 80 % in terms of accuracy using the original English questions and over a 65 % using Spanish questions and machine translation systems. Our conclu-sion about the features is that a lexical, syntactic and semantic features combina-tion obtains the best result.