Reward rank: A novel approach for positioning user answers in community question answering system

Abstract

Identifying best answers from a large set of answers from different users is the important task in Community Question Answering (CQA) sites. Distinctive methods and methodologies are used in such sites for identifying best answers and ranking the quality answers. We identified two major problems in leading CQA sites. The first problem is majority of best answers identified by CQA sites are from the users who already had many followers. The second is if a question is posted and if the best answer is written by a user late after many answer, the probability of the answer getting ranked as top answer with high rank is very less. This paper introduces RewardRank algorithm to rank the answers provided by users in CQA sites. The RewardRank algorithm treats users with more followers and less followers equally. Also, the answer written at various time intervals by different users is also treated equally. On short ranking is based on only quality of answers. We compare the RewardRank method with existing ranking methods. The experimental results demonstrate that RewardRank is the best method in ranking CQA sites.