Knowledge Graph-Based Recommender Systems to Mitigate Data Sparsity: A Systematic Literature Review

Recommender systems (RSs) have become important tools in the modern lifestyle; they have been integrated into all domains, spanning from entertainment (music, films, etc.) to more sensitive fields such as security and health care. Their success does not mean that they are ideal or flawless; quite the opposite, RSs suffer from plenty of drawbacks and challenges that need to be resolved. Data sparsity is a common problem in recommender systems; it has been of top interest among researchers. Numerous approaches from different perspectives have been proposed to mitigate it, including knowledge graphs (KGs), which quickly gained popularity due to the rich semantics residing in their components. In this paper, we will conduct a systematic literature review to explore and analyze in depth the existing contributions. Our work focuses on investigating the effectiveness of KGs to mitigate the data sparsity in RSs by discovering the techniques used, understanding how KGs are exploited, and what type of knowledge is extracted from them, besides studying their evaluation measures and discussing future directions that can strengthen the application of KGs to mitigate data sparsity in recommender systems.