With the increasing popularity of academic social networks, many users join more than one network to benefit from their unique features. However, matching the profiles of a user, despite being crucial for data verification and update synchronization, is challenging due to the differences in profile structures across different networks. In this paper, we propose an academic profile-matching approach that utilizes an Academic Knowledge Graph (AKG) to overcome the diversity problem in profile structures. Our approach includes three components: (1) candidate profile generation, which retrieves related profiles from the target network based on name similarity to the source profile; (2) profile enrichment, which uses AKG to discover relations between the attributes of the source and target profiles; and (3) profile matching, which selects one candidate as a matched profile. Through experiments on real-world datasets, we demonstrate that the proposed approach is effective in matching academic profiles across different networks, outperforming state-of-the-art baselines.

Keywords