Bringing knowledge into recommender systems

► A recommender system that composes a collaborative system is presented. ► A solution to put user knowledge into recommendations is shown. ► A simulation is performed and results indicate the effectiveness of the approach. Recommender systems are largely used nowadays to support collaborative tasks. However, it is important to consider each user's knowledge of the system for the recommended subject. In this paper we describe the use of user knowledge to improve the recommender system of the Business Process Cooperative Editor (BPCE), a collaborative business process modeling tool. We use the concept of the Knowledge Vector, developed in a previous work on collaborative navigation, to factor user knowledge into recommendations. We present Knowledge Vectors and how they are applied to the editor. A simulation to evaluate the effectiveness of the editor's new recommender system is presented.