An adaptive user interest hierarchy for filtering news

The main objective of this research is to develop an adaptive user interest hierarchy for learning static, as well as drifting user interests. In this world of information, user information needs change over time. These changes can be captured into a profile through continuous incremental learning. The idea is to recommend interesting articles to the user based on previous increment learning of his interests. This model uses bigrams to build up the user interest-learning model. This learning model consists of a set of categories, generated from continuous learning of the user's interests. It is based on the idea that terms occurring within some distance in a document are related to each other in some way. These terms can be kept in a single category if they are occurring together frequently enough. This model uses explicit feedback from the user. The system shows competence in learning static and drifting user-interests.