My research project is focused on tensor factorization technique applied to context recommender systems. I have data of user, item, movie and context- for example, weather, user's company when watching the film, location etc.

For example: user 101 watches the movie 'The Matrix' at home, with his girlfriend, and the sky is clear- leading to a rating is 4/5. My task is to understand if the context, and which context, influenced the rating of the user. Referring to the example, if the user watches the film in a different location (e.g. cinema), or with a different person, will the ratings differ? I need a technique that will enable me to answer these question- any suggestions?