Modeling temporal behavior to identify potential experts in question answering communities

Question answering (Q&A) communities are becoming important repositories of crowd-generated knowledge. The success of these communities mainly depends on the contribution of experts, who provide a significant number of high quality answers. Identifying these experts as soon as they participate in a community enables the community managers to nurture and retain experts. However, there is a great challenge to complete this task because lack of enough activities during users’ early participation. To take full advantage of users’ limited activities, we study the evolution of users’ temporal behavior that indicates deeper insights of the activities, both the absolute view and the relative view. Based on our analysis, we propose a Temporal Behavior Model to identify potential experts. Experiments on a large online Q&A community prove that our model can be combined with previous researches to improve the identification performance even further.