Automatic Question Tagging Using Multi-label Classification in Community Question Answering Sites

Community question answering sites facilitate their users to ask question and get answer from other users who has the knowledge to answer the questions. CQA sites uses tags to categorize questions into several topics so that users can easily find out the questions useful to them and related to their field of expertise. The task of annotating tags to question cannot be entirely left on users because sometimes it may happens that user who asks the question do not know the accurate tags for the question being asked. Questions with correct and accurate tags are informative and likely to get good answers. In this paper, we present a model to automatically provide tags to a question asked. We considered the problem of annotating tags to a question as a multi-label classification task. Multi-label classification is implemented to formulate question tagging as it allows adding multiple labels to a single instance. To evaluate the presented model, a set of evaluation parameters such as accuracy, hamming loss, and zero-one loss are used. The obtained results are remarkable in automatic question tagging.