MovieLens DataBase Training Data Classification Model Profile Manager Prediction Model
Demographic Profile Age Gender Occupation Genres 20 Male student Adventure, Fantasy, Children 21 Female student Fantasy, Sci-Fi, Children 32 Male Engineer Adventure, Action 33 Female Engineer Adventure, Sci-Fi, Action 34 Male Doctor Sci-Fi, War, Action 35 Female Doctor Adventure, Sci-Fi, Action 31 Female Technician Mystery, Sci-Fi, Thriller 36 Male Technician Action, Adventure, Romance, War, Sci-Fi 28 Male Entertainment Adventure, Sci-Fi, War, Action 27 Female Entertainment Adventure, Comedy, Action 32 Female Marketing Action, Adventure, Romance, Sci-Fi, War 37 Male Marketing Sci-Fi, Thriller, Action
AGE GENDER Occupation Movie Recommendations Recommender Engine User Preference Inferences
Personalized Movie Advertisements to Targeted users
Current approach to Advertisements
Rely on spamming to let users know about a new movie.
Use : Large user base to spam to
Drawbacks : Leads to user frustration
Advertises movies to users based on user profile.
Uses data mining techniques to determine the best movies to recommend to users.
Rating Occupation Movie Recommendations and Advertisements Recommender Engine/ Affinity Analysis Genre Specific User Rating Genre Movie Name
Set of rules associating Highly rated movies from different genres.
If highly rated movies of Genre A andB are associated with a highly rated movie of genre C, a user who rates movies of Genre A and B highly can be recommended a movie of Genre C with a high confidence of it being rated high.
Eg) Face off_4.5, Babel_5 => Sweeny Todd_4.
Limitations of the Approach:
Very popular movie will affect the results.
Eg) Matrix, Star Wars are always rated high and are not statistically significant