Diversification in session-based news recommender systems

Recommender systems are widely applied in digital platforms such as news websites to personalize services based on user preferences. In most of users anonymous and the only available data is sequences items sessions. Due this, typical collaborative filtering methods, which highly many applications, not effective recommendations. this context, session-based recommenders able recommend next given sequence previous active session. Neighborhood-based has been shown be compared more sophisticated approaches. study we propose scenarios make these recommender diversity-aware address filter bubble phenomenon. The phenomenon a common concern recommendation it occurs when system narrows information deprives diverse information. results applying proposed show that diversification improve diversity measures four datasets.