Topic model-based recommendation system for media re-creation service

This paper proposes a topic model-based recommendation system that predicts a user's rating score of a searched media unit for media re-creation service. The purpose of the proposed recommender is to perform a context-aware recommendation of suitable media units for a user's purpose and context. For this purpose, it uses a topic vector based on the metadata of the media to compare the similarity of subject of recreated and re-creating media. In the experiment to validate the concept of the proposed recommender, the efficiency of recommendation was improved while indicating the MAE (Mean Absolute Error) is lowered without significant time consumption.