RoadMap for building recommender system conducting requirements engineering

Abstract

Due to the fact that the incorrect definition of requirements results in unexpected cost, system unreliability and not granting users expectations, so the question that “which requirements should be taken into consideration in that domain?” is raised when building recommender systems. Furthermore, because recommender systems are used in various domains, there are very challenges when building them. So we are obliged to conduct requirements engineering which through it, better understanding of the system and requirements control to be achieved. So in this paper the process of determination of recommender system requirements in various domains has been analyzed and engineered, and in the end, a methodology for requirements engineering in recommender systems has been offered. To do this, we implemented one of the existent recommender systems, and at the next step, we built the same recommender system based on requirements engineering. In the end, we evaluated and compared the results of two recommender systems, one of them was built with and the other one without conducting requirements engineering by the use of “Precision”, “Recall” and “Response Time” metrics. While doing this study, a “RoadMap” and “Methodology” have been achieved, which we used them in another recommender system that the results of its evaluation has been presented too.