From bad to good: an investigation of question quality and transformation

Social question answering (SQA) services are a popular way for people to exchange information. Unfortunately, the quality of information exchanged can be variable and few studies focus on the quality of questions asked. To address this, we explored the influence of textual features on question quality based on 126 questions taken from five different categories of Yahoo! Answers labeled as \"Bad\" by human assessors and then revised to be \"Good\" by them. Findings indicate significant differences between the means of each feature before and after revision, suggesting the potential for an automated system that could flag questions of poor quality. In addition, by exploring the relationship between features contributing to good quality questions, we suggest a simple set of strategies askers can take when writing a question in order to improve its chances of receiving a satisfactory answer.