Learning thread reply structure on patient forums

The thread reply structure on patient forums is important for users and automated techniques to understand the discussion content and search information effectively. However, most online patient forums only have partially labeled structures. In patient forums, the discussions by patients and caregivers contain abundance of person references, which provide strong indication of the thread reply structure. In this paper, we propose using person reference resolution, combined with a statistical machine learning model, to learn the unknown thread structure on patient forums. Our preliminary performance evaluation has verified the effectiveness of the proposed approaches.