用户历史播放记录数据集
收藏资源简介:
该数据集收集了单一用户在Last.fm平台上大约15年的历史播放记录,包含超过90,000次播放和约14,000首独特曲目。数据集由用户时间上下文数据构成,每条记录代表用户在特定时刻收听的特定音乐描述符。音乐描述符包括社区贡献的Last.fm标签和Spotify音频特征。数据集的创建过程涉及从Last.fm和Spotify API下载用户播放历史和音频特征数据,并对这些数据进行预处理和聚合,以生成最终的音乐时刻数据集。该数据集旨在解决音乐推荐系统可解释性差的问题,通过分析单个用户的音乐收听习惯和上下文,提供更具针对性的推荐。
This dataset collects approximately 15 years of historical play records from a single user on the Last.fm platform, containing over 90,000 plays and approximately 14,000 unique tracks. Composed of user temporal context data, each record in the dataset represents specific music descriptors associated with the user's listening behavior at a particular moment. The music descriptors include community-contributed Last.fm tags and Spotify audio features. The dataset creation process involves downloading user play history and audio feature data from the Last.fm and Spotify APIs, followed by preprocessing and aggregation of these data to generate the final music moment dataset. This dataset aims to address the poor interpretability issue of music recommendation systems, by analyzing an individual user's music listening habits and context to provide more targeted recommendations.

- 1User-centric Music Recommendations西班牙卡斯蒂利亚-拉曼查大学计算机系统系、澳大利亚莫纳什大学信息技术学院 · 2025年



