DEVELOPMENT OF A COMMUNITY QUESTION ANSWERING SYSTEM

Abstract

A Community question answering (CQA) system enables people to post questions, comments and answers in various domains online while others reply with answers or their opinions about the posted question or comment. In this way, users can obtain specific answers to their questions, instead of searching through the large volume of information available on the web. The comments and subsequent comments on a topic create a valuable source of information. However, it takes effort to go through all possible answers and select which one is the most accurate for a specific question. Hence, extracting specific answer(s) to a question becomes vital to avoid reading every comment in the forum. In this thesis, a community question answering system was developed. Combined techniques of pattern-based and automatic query approach were used on classified questions to enhance the provision of accurate answers to such questions. Dynamic pattern was constructed from the query in relation to the extracted answer passages. In discovering questions without answers, a graph-based method, Kullback–Leibler (KL) divergence was used to identify existing related questions, and model authorship of question that were afterwards ranked. The source data for the research work were crawled from ResearchGate, a social networking website for scientists and researchers to collaborate and exchange ideas, ask and give answers to questions. The system was implemented using Java programming language, Java development kit and Java virtual machine which made the java program capable of executing on any platform. Moreover, the results of the answer extraction for questions were evaluated using precision, recall, accuracy and F-measure. Experimental results for questions crawled from ResearchGate forums showed that this method was able to isolate answers for questions. Finally, the evaluation of the model reveals high performance accuracy and precision.