Migrated Question Prediction on StackExchange

Abstract

Stack Exchange (SE) is a network of popular Community based Question and Answering (CQA) websites. Each SE Q&A website is created to address questions on specific user interest or domain. However, often users post questions on SE websites that do not match the domain of the website. Such questions are considered as Off-topic for the origin site. Off-topic questions must be detected and migrated to more appropriate On-Topic site in SE network. Off-topic questions are migrated manually to other sites by moderators or experts users (through a voting process). The process of migrating questions from one site to other is known as Question Migration. We study migrated questions on SE Q&A website. We identify several distinguishing features of migrated questions and propose a machine learning based framework to predict migrating questions. Effectiveness of proposed model is tested on five SE Q&A sites. Experimental results demonstrate that the proposed model is effective (maximum accuracy of 73%) in predicting migrating questions.