Key points are not available for this paper at this time.
The ever-increasing amount of research articles that need to be accessed has led to the development of research paper recommendation systems. Many existing algorithms merely base their suggestions on the metadata of publications, such as author names and keywords. This, however, does not reflect the context of the study and may lead to recommendations that are not accurate. To address this, we propose an innovative approach utilizing GraphSAGE within a knowledge graph which is partitioned into author, domain, and venue communities. Additionally, user intentions, inferred from user interactions with papers, play a pivotal role in tailoring suggestions. This results in a more varied and comprehensive understanding of the research landscape.
Anil et al. (Fri,) studied this question.