The Open Cybernetics & Systemics Journal

2014, 8 : 1033-1037
Published online 2014 December 31. DOI: 10.2174/1874110X01408011033
Publisher ID: TOCSJ-8-1033

Question Retrieval Based on Probabilistic Latent Semantic Analysis in Q & A Community

Chengfang Tan , Hong Li , Yundong Liu and Zhenggao Pan
School of Information Engineering, Suzhou University, Suzhou 234000, Anhui, China; and Intelligent Information Processing Lab, Suzhou University, Suzhou 234000, Anhui, China.

ABSTRACT

With the increasingly popularity of Q & A Community, it has become an important means for people to retrieve question from question library to find the answer. Similarity calculation is the core issue in Q & A community, and the appropriate calculation method is the key factor that affects the quality of question retrieval. This paper proposes a retrieval method based on PLSA model. Firstly, we modelled the question library, and got the probability distribution of "question document –latent semantic -word". Secondly, we calculated the semantic similarity between questions and classify them. Finally, based on user retrieval content, we calculated the similarity between question documents and query, then the query results will be returned to the user in descending order according to the value. Compared with other similarity calculation methods that use VSM, HNS and SD, the experimental results show that this proposed method has a high precision rate.

Keywords:

Probabilistic latent semantic analysis, , , , Q & A community, semantic analysis, similarity calculation.