MT_Paul_Foeckler

Concept and Implementation of a Group - Specific Recommender System Based on Movie Ratings and Motives

Description

The huge amount of movies plus corresponding ratings and recommendations available makes it fairly difficult to choose a movie, which possibly fits best personal desire and interest. One can hardly reckon if people who recommend movies are like-minded or the same kind of cinemagoer as oneself.

There are many systems dealing with recommendation of movies. Traditionally, those systems are using content-based, collaborative, and hybrid methods for recommendation and mostly dealing with web-based applications that have two types of entities, users and items. In content-based recommendation one tries to recommend items similar to those a given user has liked in the past, whereas in collaborative recommendation one identifies users whose tastes are similar to those of the given user and recommends items they have liked. Hybrid systems combine these two methods. Disadvantages of these techniques are limitations of user preferences through feature-based descriptions and simple ratings of movies. We believe that the user‚s attitude according to a specific domain should be centralized to create an appropriate profile for recommendations.

In this work we follow a model-based approach in order to group people into clusters with similar profiles and create group-specific recommendations based on ratings of users in the same cluster. Here, user preferences are composed of motives of cinema-going and additional ratings of specific movies, which provide a unique profile in the domain of motion picture.

The work is prototypical realised through a web-based application, where users are classified after going through steps of registration. Additionally access should be warranted by mobile phones to get recommendations and provide ratings every time and place desired.

People

Tom Gross, Supervisor
Thorsten Hennig - Thurau, Supervisor
Paul Foeckler, Master's candidate