Augmenting a Content-Based Recommender System with Tags for Cultural Heritage Personalization

Cultural heritage personalization and Web 2.0 joint researchefforts have recently emerged in the attempt to build social and collabora-tive approaches to solve the problem of filtering content in the context of art museums. One way to tackle the problem of recommending artifactsto visitors is to take into account not only the official textual descrip-tions, but also the user-generated content, namely the tags, which visitorscould use to freely annotate relevant works. The main contribution of thepaper is a strategy that enable a content-based recommender system toinfer user interests by using machine learning techniques both on staticcontent and tags. Experiments were carried out by involving real userswho annotated paintings from the Vatican picture-gallery. The main out-come is an improvement in the predictive accuracy of the tag-augmentedrecommender system compared to a pure content-based approach