Is explanation the cure? A human-centered framework for explainable recommender systems

Why am I seeing this recommendation and not something else? In a world where algorithms quietly shape what we see online, this dissertation asks whether explanations can help people better understand the recommender systems influencing their choices. Through a human-centered framework and three empirical studies, it examines for whom, in what ways, and under what conditions such explanations can make recommender systems more comprehensible, effective, and inclusive. The findings show that explanations can improve users’ understanding, but their effectiveness depends on how they are designed, whether users notice and engage with them, and on differences in users’ AI competence. Rather than treating explainability as a simple cure-all, this dissertation presents it as a human-centered challenge: designing explainable recommender systems that support meaningful engagement in AI-shaped communication environments.