Consumer preference disaggregation based on online reviews to support new energy automobile purchase decision
Chen Ying, Xingli Wu, Huchang Liao, Gang Kou · Procedia Computer Science · 2023
In recent years, new energy automobiles have developed rapidly and become the choice of more and more consumers. At the same time, a vast number of online reviews of automobiles with inferred consumer preferences emerge on platforms. There is an urgent need to mine useful information from online reviews to support consumer purchase decisions. In this study, we propose a multi-criteria product recommendation method that considers consumer’ preferences and risk psychology estimated from online reviews. Firstly, a sentiment analysis method is developed to mine product performance under different attributes from text reviews. Secondly, a preference model with unknown preference parameters is predefined based on the multi-attribute value theory to represent consumers’ value systems for purchase decision. On this basis, we propose a preference disaggregation analysis method that combines the prospect theory. Consumers’ specific preference models are estimated from online reviews, which are then utilized to measure product performance and generate product recommendations. Finally, the validity of the proposed method is demonstrated by a case study of new energy automobile ranking from Autohome.