Emotion-Annotated Podcast Dataset for Context-Aware Recommendation Research

This dataset comprises a collection of podcast items enriched with emotion annotations and contextual metadata to support research in context-aware and emotion-conditioned recommender systems. The dataset includes podcast metadata (e.g., title, category, duration), user interaction logs, and inferred or annotated emotional states associated with listening sessions. It is designed to facilitate experimentation in personalized recommendation, affective computing, and user behavior modeling in audio-based content platforms. Podcast data were collected using the YouTube Data API v3, followed by preprocessing and annotation procedures to ensure consistency and usability. This dataset can be used for benchmarking recommendation algorithms, particularly those incorporating emotional and contextual signals.