User preferences discovery using fuzzy models

User preferences discovery aims to learn the patterns of user preferences for various services or items such as movies. Preferences discovery is essential to the development of intelligent personalization applications. Based on decision and utility theories, traditional approaches to preferences discovery explicitly query users about the behavior of value function, or utility of every outcome with respect to each decision criterion. Consequently, these approaches are generally error-prone and labor intensive. Although implicit elicitation approaches have been proposed to address the above limitations, extent approaches largely ignore multi-valued nature of item features and uncertainty associated with item features and user preferences. To address uncertainty due to vagueness and imprecision, this research proposed a general framework for preferences discovery based on fuzzy set theories. In addition, new fuzzy models were created for preferences discovery and representation. Further, an algorithm was developed to predict user preferences with uncertainty, and visualization of item features, user feedback, and the discovered preferences helped improve the interpretation of the discovered knowledge. The results of the simulation evaluation using a benchmark movie dataset revealed that the proposed preference discovery method: (1) doubled the accuracy of preference discovery as compared to random prediction; and (2) outperformed conventional techniques in making movie recommendation. These findings suggest that fuzzy models are effective for preferences patterns discovery, and personalized recommendation application.