Mining FAQ from forum threads: theoretical framework

Abstract

Frequently Asked Questions (FAQ)'s tag is becoming more popular on websites. Research activities have been concentrated on its retrieval rather than construction. FAQ construction can be achieved using a number of sources. Presently, it is mostly done manually by help desk staff and this tends to make it static in nature. In this paper, a comprehensive review of various components that can guarantee effective mining of FAQ from forum threads is presented. The components encompass pre-processing, mining of questions, mining of answers and mining of the FAQ. Besides the general idea and concept, we discuss the strengths and limitations of the various techniques used in these components. In fact, the following questions are addressed in the review. What kind of pre-processing technique is needed for mining FAQ from forum? What are the recent techniques for mining questions from forum threads? What approaches are currently dominating answer retrieval from forum threads? How can we cluster out FAQ from question and answer database?