Data mining for mathematical question answering community

Abstract

Question Answering (QA) communities such as Yahoo! Answers and Baidu Zhidao are currently very popular with millions of users. The QA communities are particularly useful for the educational domain such as mathematics. Similar to traditional QA communities, a mathematical QA community should also allow users to search, ask, answer and discover mathematical questions. However, as mathematical formulas are highly symbolic and structured, it is challenging to develop such a mathematical QA community. In this research, we aim to propose efficient and effective techniques for supporting the ”search, ask, answer and discover” framework for a mathematical QA community. In particular, we focus on investigating different data mining techniques for mathematical question search, mathematical question topic classification and human expert finding. Mathematical question search will help retrieve a set of similar mathematical problems together with the answers posted by other users. Mathematical question topic classification will help recommend the possible topics of user posted questions. Human expert finding will help find a list of experts who are most likely able to answer a posted question according to their expertise.