This is a neural model for Community Question Answering task in SemEval2016. The task has three parts - A, B, and C. More detaisl can be found about the task here: http://alt.qcri.org/semeval2016/task3/ For task A, we have used a model proposed by Wang et al. (2016) - Sentence similarity learning by lexical decomposition and composition. For task B, we have used TF-IDF based method. For task C, we have used Task A and Task B.