A Hybrid Approach for Spatial Web Personalization

In the context of Web personalization, Markov chains have been recently proposed to model user’s navigational trails, in order to infer user preference and predict future visits through computation of transitional probabilities. Based on these principles, the research introduced in this paper develops a hybrid Web personalization approach that applies k-order Markov chains towards an integration of spatial proximity and semantic similarity for the manipulation of geographical data on the Web. This framework personalizes Web navigational experiences over spatial entities embedded in Web documents. A reinforcement process is also introduced to evaluate and adapt interactions between the user and the Web on the basis of user’s relevance feedbacks. An illustrative case study applied to spatial information available on the Web exemplifies our approach.