Question And Answer Recommendation In Question Answering Communities

User-Interactive Question Answering (QA) communities such as Yahoo! Answers and Baidu Zhidao are growing in popularity. However, as these QA sites always have thousands of new questions posted daily, it is difficult for users to find the questions that interest them. Consequently, this may delay the answering of new questions. Meanwhile, as the number of candidate answers increasing, the asker may have difficulty in choosing the best one from those with uneven quality.In this paper, we study the question and answer recommendation mechanism to help the asker and answerer seek information and enhance the knowledge sharing activities within question answering communities. Question recommendation techniques help users locate interesting questions and expedite the answering of new questions. We believe users may select questions according to their own preferences. Thus in this paper, we adopt the Probabilistic Latent Semantic Analysis (PLSA) model to present users' preference distribution according to users' previous answering history, thus generating recommending question lists. To help askers find the best answers, answer recommendation techniques rank candidate answers automatically. The recommendation is conducted based on the content similarity and user authority. We summarize the relationship between users as link structure, and adopt the PageRank algorithm for estimating user authority. The similarity calculation considers both the similarity between questions and answers, and that between askers and answers. The experimental results show our topic-modeling based question recommendation approach can capture users' preference and recommend questions effectively. Experimental results of answer recommendation conclude the effectiveness of considering both the similarity between questions and answers, and that between askers and answers.