Context-aware Music Recommendation Based on Word2Vec

Musical contextual factors can affect the users’ preferences for music greatly, so it is necessary to take the users’ current contextual factors into account when making music recommendations to the user. However, we face two critical challenges: how to get the users’ contextual information and how to integrate the contextual information into the recommender systems. In this paper, a method is proposed for extracting contextual factors using the word2vec algorithm which is concerned to context-aware information. The neural network model is used to obtain distributed representation of the music pieces. According to the learned distributed representation, the users’ long-term and the short-term music preferences can be predicted. Then, a word embedding model is presented, which can incorporate the contextual information into the recommender system with the cosine similarity between the representations of tracks and the users. Finally, the most similar tracks are recommended to the target users.

Word2vec, music recommender system, word embedding, context-aware

Yijie zhou, Pei Tian, Context-aware Music Recommendation Based on Word2Vec. 2019 International Computer Science and Applications Conference (ICSAC 2019). 2019: 50-54.