1) The #nowplaying-RS dataset consolidates different data types from Twitter and Spotify, including item content, user contexts, and timestamps, to build context-aware music recommender systems. 2) It contains over 11.6 million listening events from 140k users for 350k tracks by 45k artists and with 45k hashtags and sentiment scores. 3) The paper demonstrates context-aware recommendation and next-song recommendation using the dataset, evaluating methods like factorization machines and reporting results using mean reciprocal rank.