Selecting Best Answer: An Empirical Analysis on Community Question Answering Sites

Abstract

A community question answering (CQA) site is a well-known online community, where user interacts on a wide variety of topics. To the best of our knowledge, the selection of a best answer for the question asked on the CQA site is done manually, which is traditional and tedious. In this paper, a model is developed for selecting best answer for the question asked on the CQA site. Instead of taking data related to question–answer only into account as done in manual process, this model takes both question-answer and answerers’ data into account, which gives an insight view into the answers given by the experts that is more likely to be selected as the best answer. The presented approach analyzes StackOverflow Q&A posts with at least five answers to extract features for pattern identification using which the best answer is selected for the asked questions based on topic modeling and classifier. To evaluate correctness of the proposed model, a set of parameters are used, such as Receiver Operating Characteristics Area Under Curve, Precision Recall Area Under Curve, Gmean, and Accuracy. Results show that the proposed model is effective in predicting the best answer.