| 摘要 |
A computerized system and an associated computer-implemented method for the analysis of user activity and preparation of the data for a music recommender in a social network. The history of actions is analyzed in multiple dimensions in order to mine collaborative correlations, temporal correlations and overall ranking. The results of the analysis are exported in a form of a taste graph, which is then used to generate on-line music recommendations. The taste graph captures relations between different entities pertaining to music (users, tracks, artists, etc.) and it consists of the following main parts: user preferences, track similarities, artist similarities, artists' works and demography profiles. Each part of the taste graph is created using a separate algorithm. The recommendations are generated based on the composed stochastic graph structure using a random walk algorithm. |
