A Reqirements Engineering Approach to Attribute Selection in a Recommendation System

摘要

Attributes play an important role in recommendation systems. This paper presents a method to enhance the result of a recommendation system through attribute selection based on user goals. With a limited amount of data, a recommendation system suffers from sparsity problems and cold start problems in collaborative filtering. In order to utilize such data, our method seek to fmd a set of suitable attributes using goal-driven requirements analysis to provide more timely information of user preferences for the current needs of a user. The process of goal-driven analysis is aided by the Analytic Hierarchy Process in attribute selection. The experiments show that this method is feasible and derives promising results.