Dependency Graph Based Detection of Semantically Equivalent Questions in Online Forums

Introduction

The online question-answer forums like Quora (https://stackoverflow.com/) have gained considerable popularity amongst the online user community. The number of users as well as the number of questions asked on these online Question-Answer Forums is increasing each day. In order to provide efficient search of appropriate answers, the forums, in general, ask the users to search the forum for existing answers prior to posting a new question. However, a question may be framed and asked in different manner by different users and therefore, searching for similar questions is not an easy task. This results in the addition of semantically equivalent questions on the online user forums.

Two questions have been defined to be semantically equivalent if exactly the same answer can be used to effectively answer these questions (Jiwoon Jeon, Croft & Lee, 2005; Bogdanova, 2005). For instance, the following two examples (Jiwoon Jeon, Croft & Lee, 2005) intend to ask the same question:

• •

  Can I attach a 5 megabyte file in my email?

• •

  Sending big movie files to my friends over the net by email

As another example included in a sample dataset provided by Quora (https://data.quora.com/First-Quora-Dataset-Release-Question-Pairs). The dataset has been released to obtain an efficient solution for this challenging problem.