As recommender systems transition from “black boxes” to explainable models, assessing the quality of their explanations has become a critical research challenge. A systematic literature review has been performed to analyse how data-based explanation quality is evaluated in recent research (2021-2026). Findings reveal a significant reliance on system-oriented methods and metrics, while direct human-centred evaluation remains underrepresented.
This work is licensed under a Creative Commons Attribution 4.0 International License.