Towards Personal Explanations for Recommender Systems: A Study on the Impact of Familiarity and Urgency

Recommender systems shape online experiences, but their complexity often hinders user understanding. Explainable recommender systems address this by providing meaningful explanations. Our research explores how explanations can be tailored to individual users and contexts. Through an online survey with 16 participants, we investigated how different levels of domain familiarity and decision-making urgency influence the perception of four types of explanations and their corresponding qualities. Findings shows that transparency and completeness are valued across all explanation types, regardless of users' familiarity levels. Moroever, of the four studied explanation types, behavioral explanation while being the one already available across multiple commercial RecSys underperform when compared with the other three types. Urgency influences explanation quality, with completeness favored in non-urgent situations and simplicity becoming more important as urgency rises. These results, contribute to the general understanding of how human characteristics like familiarity and urgency can influence the design of explanations in recommender systems.