Global academic exchange and cooperation have become an increasing trend in both academia and industry, but how to quickly and effectively identify potential partners is becoming an urgent problem. This paper proposes a link prediction-based model to help researchers identify partners from a large collection of academic articles in a given technological area. We initially construct a co-authorship network, and take a series of indices based on network and similarity of researchers into consideration. A fitting model of link prediction is then established, in which logistic regression analysis is involved. An empirical study on four journals of informetrics is conducted to demonstrate the reliability of the proposed method.
Author(s): Lu Huang, Yihe Zhu, Yi Zhang, Xiao Zhou, Xiang Jia
Organization(s): Beijing Institute of Technology
Source: 2018 Portland International Conference on Management of Engineering and Technology (PICMET)