Coauthorship networks and academic literature recommendation

Recommender systems are increasingly touted as an indispensable service of many online stores and websites. Most existing recommendation techniques typically rely on users’ historical, long-term interest profiles, derived either explicitly from users’ preference ratings or implicitly from their purchasing/browsing history, to arrive at recommendation decisions. In this study, we propose a coauthorship network-based, task-focused literature recommendation technique to meet users’ information need specific to a task under investigation and develop three different schemes for estimating the closeness between scholars based on their coauthoring relationships. We empirically evaluate the proposed coauthorship network-based technique. The evaluation results suggest that our proposed technique outperforms the author-based technique across various degrees of content coherence in task profiles. The proposed technique is more effective than the content-based technique when task profiles specified by users are similar in their contents but is less effective otherwise. We further develop a hybrid method that switches between the coauthorship network-based and content-based techniques on the basis of the content coherence of a task profile. It achieves comparable or better recommendation effectiveness, when compared with the pure coauthorship network-based and content-based techniques.