Cross-domain Scientific Collaborations Prediction Using Citation

Cross-domain Scientific Collaborations have promoted rapid development of science and generated many innovative breakthroughs. However, predicting cross-domain scientific collaboration problem is rarely studied in academic research. Moreover, collaboration recommendation methods within single domain cannot be directly utilized for solving cross-domain problems, because there are topic skewness and sparse connections challenges. In this paper, we propose a Hybrid Graph Model, which combines both explicit co-author relationships and implicit co-citation relationships together to construct graph, then Random Walks with Restarts concept is used to measure and rank relatedness between nodes. Because co-citations appear in both source domain and target domain, they represent the topics which can be shared across domains. In this way, topic skewness problem is solved much cheaper and effectively compared with probabilistic topic models. In addition, co-citation relationship solves sparse connection problem by mining more potential connections between authors. However, few previous works use citation information for scientific collaboration recommendations. Finally, we compare the performances of Hybrid Graph Model with some baseline approaches on large publication data sets from different domains. The experiments show that Hybrid Graph Model outperforms comparison methods on several recommendation metrics. Moreover, citation information has been demonstrated to be very helpful for scientific collaboration recommendations.