600 likes | 1.45k Views
Movie Recommendation based on movie feature. Annan Wei. Outline. Movie Recommendation System Data Source Algorithm UI. Movie Recommendation System. Generate random movies list from database User can choose the movie they prefer Calculate User Feature Value
E N D
Outline • Movie Recommendation System • Data Source • Algorithm • UI
Movie Recommendation System • Generate random movies list from database • User can choose the movie they prefer • Calculate User Feature Value • Calculate the distance between User Feature Value and Movie Feature Value • Recommend another movie they might like
Movie Recommendation System • Collaborative filtering The user will be recommended items that people similar tastes and preferences liked in the past • Content-Based Filtering The user will be recommended items similar to the ones the user preferred in the past • Model-Based Filtering
Movie Recommendation System User selection UI Profile Learner Profile Learner List of recommend movie Filtering Component Content Analyzer Movie Descriptions feedback Active user IMDB
Data Source 5,826,213 personalities 52 million registered users 2,841,405titles Alexa Top 50 site Internet Movie Database Allows users to download the entire database locally (at http://www.imdb.com/interfaces) to their own machines for free
Data Source Figure 1: The table of Movie Title Figure2: The table of Name(e.g. actor's name)
Assumption • Assumption 1. User select this movie by movie’s feature. • Assumption 2. User want to get the movie with high rating.
Calculate Movie Feature Value "Action" "Adult" "Adventure" "Animation" "Biography" "Comedy" "Commercial" "Crime" "Documentary" "Drama" "Experimental" "Family" "Fantasy" "Film-Noir“ … … • Create Movie_Feature tables • in database • Search each movie in the database • Featurevector: 31-dimensional 31 genres
Calculate User Feature Value • Feature vector: 31-dimensional • Finally feature value • Create User_Movie_Like Table
Recommend Movie • K-Nearest Neighbor Algorithm • K=20
Recommend Movie • According assumption 3. User want to get the movie with high rating. • The formula for calculating the Movie weighted Rating gives a true Bayesian estimate: • WR = (v ÷ (v+m)) × R + (m ÷ (v+m)) × C R = average for the movie (mean) = (Rating) v = number of votes for the movie = (votes) m = minimum votes required (currently 1000) C = the mean vote across the whole report (currently 6.9)