Expert finding for web-based question-answering system

Abstract

Large numbers of Question-Answering (QA) communities are available on the World Wide Web such as Yahoo! Answer and Answers.com which are popular and support millions of users. An educational QA community is a QA community that focuses on educational subjects like mathematics and programming languages. In this Project, we will particularly look at a QA community which specialises in the mathematics subject. As a QA community which specialises in mathematics, it will have higher chance in covering a wide variety of mathematics questions compared to a general QA community. Thus, it will eventually attract users who especially need help in mathematics questions and also users who are expertise in mathematics questions. The QA community supports basic functions which allow users to browse, search, ask and answer questions.

In this research, we focus on building the QA community using Python with the Django framework and investigating data mining techniques used in mathematics question search and approaches in human expert finding. Python is a general purpose programming language which is very suitable in developing web applications. The goal in mathematics question search is to retrieve a set of similar questions according to user’s input query. Whereas for human expert finding, the goal is to retrieve a set of experts in a particular topic, subject or in a particular set of questions.

From the research, suitable approaches for mathematics question search and human expert finding are proposed and also integrated to the QA community (BingoQA). The proposed approaches have been evaluated and resulted in promising performance. In this report, the motivation, objectives, related work, proposed approaches and their performance evaluation will be presented. Future work in enhancing the QA community will also be discussed in this report.