Consumer Character Analysis And The Corresponding Recommendations

In order to cope with information overload, user personalization issues, e-commerce recommendation has become an important solution. But with user personalization growing and new e-commerce emerging, traditional e-commerce recommendation has been unable to meet the needs of its own development. Consumer buying behavior is determined by consumer psychology, which exist a causal relationship between them. More importantly, consumer psychology is much more stable than consumer behavior. This paper introduced the theory of consumer character into e-commerce research.Character is divided by attitudes and buying patterns; preferences are often obtained from product attributes and user behavior regularity, the paper launched two aspects of the study. First, we analyze consumer characters type from attitudes and buying patterns dimensions. In the first dimension, this article defines some attribute variables about transaction records and establishes the relationship between the consumer characters and the product properties using Bayesian network to analyze consumer character. In the second one, it defines consumer behavioral variables related with clickstream and establishes the relationship between the consumer characters and user behavior to analyze consumer character. Second, we proposed e-commerce recommendations. The attitude perspective character analysis can do Products Recommending with Bayesian network inference. The buying patterns perspective character analysis can do page recommendation.This paper select a user’s transaction records and clickstream to do research, which often surf on the Web and buy goods. Research results show:Bayesian network can well establish links between product attributes and consumer character, user behavior and consumer character. From network structures, the paper finds property variables and behavioral variables directly affected by consumer character. According to the stability of character, do commodity and page recommendation respectively. This article combines the qualitative Bayesian theory and quantitative user character, grasp accurately the consumer character and make corresponding recommendations methods. This paper provides a reference for comprehensive analysis of consumer psychology, but also a new idea to develop new recommendations methods on the base of consumer psychology.