Predicting potential responders in social Q&A based on non-QA features
Zhe Liu, Bernard J. Jim Jansen · 2014
Given the recent advancement of online social networking technologies, social question and answering has become an important venue for individuals to seek and share information. While studies have suggested the possibilities of routing questions to potential answerers for their help and the information provided, there is little analysis proposed to identify the characteristics that differentiate the possible responders from the nonresponders. In order to address such gap, in this work we present a model to predict potential responders in social Q&A using only non-QA-based attributes. We build the classifier using features from two different aspects, including: features extracted from one's social profile and style of posting. To evaluate our model, we collect over 20,000 questions posted on Wenwo, a social Q&A application based on Weibo, along with all their responders. Our experimental results over the collected dataset demonstrate the effectiveness of responder prediction based on non-QA features and proposed potential implications for system design.