Advances in scientific article recommendation systems: A systematic literature review

Abstract

The abundance of available articles highlights the growing relevance of recommendation systems. This study conducted a systematic review of the literature with the aim of identifying the state of the art in scientific article recommendation systems. Approaches were identified that use embeddings to capture semantic similarity and probabilistic matrix factorization to integrate relationships between articles and authors. The results indicate a predominance of hybrid models, using LLMs and time-aware modeling. It is concluded that, although there have been significant advances, gaps remain in the standardization of metrics and reproducibility of experiments, as well as opportunities for the development of contextually customized models.