Context-aware Content Discovery for Cultural Collections

Type of Thesis: 
Master Thesis

Recommender systems have been – and still are – the subject of extensive research and have been applied in various application domains such as e-learning, e-commerce and even entertainment. Several techniques exist and can be broadly categorized into collaborative filtering (analyzing large amounts of users’ behavior, preferences, etc.), content-based filtering (focusing on the similarities of items to recommend with respect to a user’s prior preferences), and a combination of the two dubbed hybrid recommender systems. The inclusion of context is still a challenge with vast potential. For instance, while travelling, the user wishes an emphasis on items related to the location. Another example is to provide items that are particularly relevant during the time such as the anniversaries of important civil wars or upcoming political events. The objectives of this thesis are to incorporate such context variables in the recommendation process applied to the cultural domain and to develop an adequate prototype (ideally a browser plugin), both to be evaluated. This thesis is suitable for both 1 year and 2 year MSc programs.

Background Knowledge: 
  • Good programming skills (HTML, JavaScript and CSS, as well as backend programming)
  • Conceptual Modelling
  • Knowledge of Semantic Technologies (e.g., via the Open Information Systems course) might come in handy
Technical challenges: 
  • Context modelling
  • Design and implement a recommender systems applied to a particular domain
  • Set up, conduct and interpret the results of an experiment
  • Privacy will be a notation that should be taken into account
Contact: 
Christophe Debruyne
Academic Year: 
2015-2016