Multitask Learning with Deep Neural Networks for Community Question Answering
Daniele Bonadiman, Antonio Emmanuele Uva, Alessandro Moschitti · Italian Journal of Computational Linguistics · 2017
Traditionally, these tasks have been tackled by designing systems/classifiers that target each task separately.Each classifier accepts a vector encoding a text pair (e.g., a question/question or a question/answer pair) in input by using many complex lexical syntactic or semantic features and, then, computing similarity between these A. B.C.