An Object-Event Reading Interest Model

Extracting reading interests from a user´s reading history is a significant issue of personalized text recommendation. Most previous text recommendation methods only distinguish the interested class from uninterested class, which essentially presumes there is only one angle of reading interests for a user. However a user may have multiple angles of reading interests. Different angles of reading interests indicate different principles for recommending texts. This paper firstly distinguishes the object reading interest and the event reading interest of a user, builds a model to represent the two kinds of reading interests, and then gives two match degrees to measure the closeness between a text and a reading history in terms of the two angles. Experiments demonstrate that texts can be effectively recommended in terms of the two angles.