Collaborative Filtering Method with the use of Production Rules
Amin Salih Mohammed, Єлизавета Владиславівна Мелешко, Saravana Balaji B, Serhii Semenov · 2019
this paper proposes a new collaborative filtering method with the calculation of unknown similarity coefficients between users via the application of production rules, which aims to improve the work quality of recommendation systems, to develop of production rules for the developed system. The methods used are: graph theory, the theory of algorithms, mathematical statistics, object-oriented programming, and fuzzy logic. The developed systems are new collaborative filtering method with the definition of unknown similarity coefficients between users through the application of production rules was developed, software for the implementation and testing of this method was developed, experiments on the developed software was conducted. The production rules to determine unknown similarity coefficients in recommendation systems was proposed. The new method of collaborative filtering with the application of production rules has been developed to find unknown similarity coefficients between users that may be possibly used to improve the work quality of the recommendation system. The conducted experiments showed that the developed method enhances the quality indicators of the recommendation system, such as item space coverage and the total number of predicted preferences of users.