In recent years, the huge development in information technology led to a data explosion on the web, motivating the need for powerful and efficient strategies for information retrieval. Personalized Web systems are an example used to enhance the user experience by offering tailor-made services according to his profile. Building accurate profiles representing the reel user's interests that can change in time is the major ingredient for an efficient personalization system. This work presents our approach for generating accurate and dynamic user profiles implicitly by tracking and capturing the user's interests and preferences. Moreover, we investigate techniques to improve the profiles' accuracy; through accumulating more browsing data from multiple sources, distinguishing the most relevant concepts, and also identifying the number of ontology levels in the conceptsa€? hierarchy needed to accurately represent each user's reel interests and preferences. Exploiting users' feedback, results prove feasibility and accuracy of the generated profiles.
展开▼
