A Correlation-Based Approach to Consumer Oriented Evaluation of Product Recommendation

摘要

The paper deals with the multi-attribute aggregation problem in product recommendation, which focuses on consumers' preferences and personal tastes. Specifically, an ontological structure is used for aggregating the attributes selected by consumers as their preferences; in particular this paper also takes the correlation effects between the selected attributes into account. Consequently, this paper presents two kinds of aggregation models based on ontological structure and correlation effect for comparison, and to recommend a ranking list to consumers, four ranking methods are also examined. To make an evaluation of the aggregation models and the ranking methods, a recommendation system was developed and a comparison test was conducted.