Assessing Document Relevance by Modeling Citation Networks with Probabilistic Graphs

Paper citation networks are a traditional social medium for the exchange of ideas and knowledge. In this paper we use citation networks as a mean to assess both the importance of the citations of a paper and to identify relevant papers. We addressed these problems by modeling the citation network with a probabilistic graph useful to infer unknown links among the nodes representing papers. The proposed approach has been evaluated on three real world citation network whose results proved its validity.