Movie Recommender System

What movie should you watch tonight? It's really a hard choice since there're so many movies that even scanning their brief introductions will cost us a lot of time. So we do need a personalized recommendation engine to help narrow the universe of potential films to fit our unique tastes. Fortunately, with the help of machine learning technique, it helps users to survive from enormous volume of information and provides valuable advices about what they might be looking for based on their particular information, such as profile, searching history, etc. Product recommendation on Amazon.com is one of those successful examples in this field.

As a matter of fact, there are new movies released even every day, however, comparatively few tools can hlep us organize these content and directly pick those movies that are more likely to interest us. To address this problem, we want to develop a hybrid Movie Recommender System based on Nerual Networks, which takes into consideration the kinds of a movie, the synopsis, the participants (actors, directors, scriptwriters) and the opinoin of other users as well[1], in order to privde more precise recommendations.

Generally speaking, most existing recommender systems adopt either collaborative filtering method or content-based filtering. Collaborative filtering matches persons with similar interests and provides recommendations based on this matching. The recommendation quality of this method is generally high, however, the cold-start problem is encountered when a new item is added, since it has no evaluations at the very beginning[1]. Content-based filtering method provides recommendations according to the similar items that the user has liked in the past, but it does not take into consideration the subjective attributes, like the quality of filming[1].

In our system, we use hybrid method, which combines these two approaches, in order to construct a system that can provide more precise recommendations.

Our project based on the MovieLens dataset, which is available on the GroupLens website of the University of Minnesota[2].

This data set consists of:

• 1,000,209 anonymous ratings (in form of 1-5) from 6,040 users on 3,900 movies
• Each user has rated at least 20 movies
• Simple demographic infor for the users, including age, gender, occupation, and zip code

We also collect more detailed synopsis information of each movie (director, actors and script writers) automatically by using the URLs provided by MovieLens.

Before milestone, we should have:

• Built a model for our training data;
• Implemented a naive recommendation algorithm, which only based on user rating, used to compare with our hybrid method.

[1]. Christina Christakou, Andreas Stafylopatis, "A Hybrid Movie Recommender System Based on Nerual Networks", isda, pp.500-505, 5th International Conference on Intelligent Systems Design and Applications (ISDA'05), 2005
[2]. http://www.grouplens.org/node/73
[3]. Christina C., Leonidas L., Spyros V. and Andreas S, "A Movie Recommender System Based on Semi-supervised Clustering", CIMCA-IAWTIC, Volume 2, pp.897-903 International Conference on Computational Intelligence for Modeling, Control and Automation, and International Conference on Intelligent Agents, Web Technologies and Internet Commerce (CIMCA-IAWTIC'05), 2005
[4]. http://en.wikipedia.org/wiki/Recommender system