Recommendation system for documentary classification

In the context of the NAR project “Miipa-doc”, we develop a new type of Knowledge Organization System (KOS) called Hypertagging based on the tagging of electronic documents and the principles of faceted classification. It was designed to simplify the tasks of information management for the organizations' staff. In this paper, we propose a new recommendation model and algorithm which are based on a faceted classification by level in the aim to facilitate the documents' indexing. This approach exploits the user trace indexing of his/her documents to learn about the user preferences and then to produce their recommendations. Consequently, these recommendations will provide a kind of knowledge base aiming at improving document ranking and highlight most relevant information that meeting user needs. This model is based on a statistical method called Association Rules (AR) using an Apriori algorithm to generate the recommendations.