Question Understanding and Similarity Computation Method Based on Semantic Analysis
In order to search the similar questions accurately, this paper provides a question understanding method based on the semantic chunk recognition and the question split. Based on the understanding, a quadruple-based question representation model which repr
- PDF / 1,712,660 Bytes
- 10 Pages / 439.37 x 666.142 pts Page_size
- 24 Downloads / 232 Views
Question Understanding and Similarity Computation Method Based on Semantic Analysis Xia Yan
Abstract In order to search the similar questions accurately, this paper provides a question understanding method based on the semantic chunk recognition and the question split. Based on the understanding, a quadruple-based question representation model which represents a question as a quadruple is proposed. Furthermore, in order to deal with the different wording style of different uses in the QA community, a word similarity computation method based on synonymy thesaurus in the question similarity computation is brought up. Thereby, based on the above representation model and algorithm, a question similarity computation method based on the multi-feature fusion is provided. Experimental results show that the proposed model and method are successful, and can effectively improve the performance of similar question searching. Keywords QA community Question understanding
Similarity computation
Semantic analysis
75.1 Introduction In recent years, there appears a new question and answering system—Community Question and Answering System (CQA), which is also referred as Q&A community, cooperative question answering system, and so on, for example, Baidu
X. Yan (&) School of Computer Engineering, Shenzhen Institute of Information Technology, 518029 Shenzhen, China e-mail: [email protected]@qq.com
W. Lu et al. (eds.), Proceedings of the 2012 International Conference on Information Technology and Software Engineering, Lecture Notes in Electrical Engineering 211, DOI: 10.1007/978-3-642-34522-7_75, Springer-Verlag Berlin Heidelberg 2013
699
700
X. Yan
zhidao,1 Sinaiask2 and Yahoo! Answers.3 Compared to the traditional search engine based on keyword matching, in the CQA, users will ask a question and wait for the answers, and other users who know the answer will give directly his answer to this question. So, in the CQA, people can get the answers more quickly, accurately and effectively, and they don’t need to browse lots of web pages before they find their needed answer as they usually do in the traditional search engine. And for this reason, CQA develops rapidly, and play important roles in many fields. CQA is now a research focus in the natural language processing. With the development of CQA, CQA has amassed lots of problems which have been resolved by the users. Question similarity computation is a research focus in the CQA, which can be used to find the similar question. If a user asks a question which has been resolved before (or its similar question), CQA should return the user with the answers immediately based on the similarity computation between the asked question and the questions in the database. Now, researcher usually adopt the traditional Vector Space Model (VSM) to represent a question when computing the question similarity. This may cause many defects. CQA is an open platform, which allows any user to ask the question. Different people have different wording habit and different writing style
Data Loading...
