Detecting high-quality posts in community question answering sites

Abstract

Community question answering (CQA) has become a new paradigm for seeking and sharing information. In CQA sites, users can ask and answer questions, and provide feedback (e.g., by voting or commenting) to these questions/answers. In this article, we propose the early detection of high-quality CQA questions/answers. Such detection can help discover a high-impact question that would be widely recognized by the users in these CQA sites, as well as identify a useful answer that would gain much positive feedback from site users. In particular, we view the post quality from the perspective of the voting outcome. First, our key intuition is that the voting score of an answer is strongly positively correlated with that of its question, and we verify such correlation in two real CQA data sets. Second, armed with the verified correlation, we propose a family of algorithms to jointly detecting the high-quality questions and answers soon after they are posted in the CQA sites. We conduct extensive experimental evaluations to demonstrate the effectiveness and efficiency of our approaches. Overall, our algorithms can outperform the best competitor in prediction performance, while enjoying linear scalability with respect to the total number of posts.