作者: Yanhong Zhou , Gao Cong , Bin Cui , Christian S. Jensen , Junjie Yao
DOI: 10.1109/ICDE.2009.44
关键词:
摘要: Online forums contain huge amounts of valuable user-generated content. In current forum systems, users have to passively wait for other visit the systems and read/answer their questions. The user experience question answering suffers from this arrangement. paper, we address problem "pushing" right questions persons, objective being obtain quick, high-quality answers, thus improving satisfaction. We propose a framework efficient effective routing given top-k potential experts (users) in forum, by utilizing both content structures system. First, compute expertise according system—-this is estimate probability an expert based on previous user. Specifically, design three models task, including profile-based model, thread-based cluster-based model. Second, re-rank measured structural relations among results two steps can be integrated naturally probabilistic model that computes final ranking score each Experimental show proposals are very promising.