The Impact of Explanation Characteristics on the Effects of Recommender Systems Explanations

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

There seems to be a consensus that adding explanations to AI systems, such as recommender systems, can have positive effects, such as increasing a user’s trust, the transparency of the system, or the efficiency of a decision-making process. It is to this date unclear, though, how an explanation needs to be designed to evoke these effects. To address this gap, we conducted a systematic literature survey of 180 articles. The corpus was created by first querying six literature databases with predefined keyword searches and sorting the retrieved articles by applying inclusion and exclusion criteria. We then extracted and categorized the explanation characteristics, such as their modalities and content, as well as the effects measured in user evaluations. Finally, we combine both dimensions to analyze the impact of the characteristics on the effects. The findings suggest that it is not possible to make general statements about the effects of explanations. Instead, the measured effects vary in different contexts, and the results depend on whether the evaluated explanation was compared to another explanation or to a system without explanation. Our survey identifies and highlights research gaps and promising areas for future work, such as the effects of non-local explanations on the user.