Bias in Knowledge Graphs -- an Empirical Study with Movie Recommendation and Different Language Editions of DBpedia
This work addresses bias issues in recommender systems for users relying on multilingual knowledge graphs, though it is incremental as it focuses on empirical observation rather than novel solutions.
The study tackled the problem of bias in knowledge graphs by empirically analyzing how different language editions of DBpedia affect movie recommendation systems, showing that these variations lead to differently biased systems and performance differences in specific recommendation fields.
Public knowledge graphs such as DBpedia and Wikidata have been recognized as interesting sources of background knowledge to build content-based recommender systems. They can be used to add information about the items to be recommended and links between those. While quite a few approaches for exploiting knowledge graphs have been proposed, most of them aim at optimizing the recommendation strategy while using a fixed knowledge graph. In this paper, we take a different approach, i.e., we fix the recommendation strategy and observe changes when using different underlying knowledge graphs. Particularly, we use different language editions of DBpedia. We show that the usage of different knowledge graphs does not only lead to differently biased recommender systems, but also to recommender systems that differ in performance for particular fields of recommendations.