CHIP art recommender; finding the algorithm for an optimal result
|has title::CHIP art recommender; finding the algorithm for an optimal result|
|Master:||project within::Knowledge Technology and Intelligent Internet Applications|
|Student name:||student name::Wouter Slokker|
|Second reader:||has second reader::Annette ten Teije|
|Company:||has company::VU (CHIP project)|
The CHIP Art Recommender is a Web-based rating dialog for artworks/topics to build a user profile, based on semantics-driven recommendations. The project is a collaboration between the Technical University Eindhoven, the Rijksmuseum Amsterdam and the Telematica Institute. There are many strategies in the field of recommendation, but not every strategy is suitable for this project, and not every strategy will perform equally well. During the master project I want to find a strategy that generates good recommendations to user.
Abstract KIM 2
My thesis is about finding a strategy for an art recommender. A user will rate some artworks, and based on these ratings the art recommender should come up with artworks or topics (like baroque, golden age, oil painting) that the user would also like. This reasoning could be done in many ways. In the current situation the system looks at the properties of the rated artworks, and searches for artworks with similar properties to recommend (in case of a positive rating by the user) or reject (in case of negative ratings) them. I have searched for other strategies to do this, in the field of collaborative filtering en clustering. During my presentation I will explain what how I've come to my strategies, how I tested them and how they performed.