Query-oriented citation recommendation based on network correlation

With the rapid proliferation of information technology, researchers find it more and more difficult to rapidly find appropriate reference papers for an authoring paper. Citation recommendation aims to overcome this problem by providing a list of reference papers given a query document. There exist various aspects in bibliographic literature acting as paper's scholarly roles, such as paper's content, paper's author, citation behavior, paper's topic. We argue that combining different kinds of paper's scholarly roles can enhance citation recommendation performance. Based on it, we propose a network correlation based query-oriented citation recommendation approach. We first construct a semantic network and a citation network, these two networks consist of the same vertices but different edge connection. Then we build correlations of these two networks and select the top features to calculate the semantic similarities of the query paper and scientific papers. Finally, we choose the top ranked scientific papers as the recommended citation list. When evaluating on the AAN dataset, the experimental results demonstrate the efficacy of the proposed approach.