Combining collaborative and content-based filtering using conceptual graphs

Collaborative Filtering and Content-Based Filtering are techniques used in the design of Recommender Systems that support personalization. Information that is available about the user, along with information about the collection of users on the system, can be processed in a number of ways in order to extract useful recommendations. There have been several algorithms developed, some of which we briefly introduce, which attempt to improve performance by maximizing the accuracy of their predictions. We describe a novel algorithm in which user models are represented as Conceptual Graphs and report on results obtained using the EachMovie dataset. We compare the algorithms based on the average error of prediction and standard deviation and discuss our method's strengths and advantages.