Feature-Based Question Routing in Community Question Answering Platforms

Community question answering (CQA) platforms became popular and indispensable sources of information in different domains. The success of these platforms relies heavily on the timely contribution of their expert users who would answer questions. There are many questions that remain unanswered for a long time, if not ever, on CQA platforms. In this dissertation, the problem of question routing on CQA platforms is addressed, which aims to connect experts to the right questions. We introduce and semantically classify 67 features and then train a learn to rank framework over five different CQA datasets. Results show that features based on tags, topics, user characteristics and user temporality are effective for question routing. Also the proposed approach outperforms the state-of-the-art neural matchmaking methods, that lack the interpretation of features, without compromising interpretability. The interpretation of features’ influence helps future research in this field to address the question routing problem more effectively.