Customer Consumption Preferences of B2C Website Based on Bayesian Network Model

customer reviews are the important data source to recognize customer consumption preference of online shopping site. This paper aiming at customer consumption preferences of B2C website, customer reviews of clothing products are grabbed and preprocessed to extract feature factors. The Bayesian network model of B2C customer preferences is constructed to calculate node conditional probability distribution. Sensitive factors was recognized and dynamically adjusted. The results show that: Under the influence of the parent nodes, the customers’ evaluation of each characteristic factor variable’s subnode has a higher probability of “moderate” and “good”. The probability change of the overall evaluation node and sensitivity factors has a positive impact on the probability of other factors. Customer consumption preferences can be judged and predicted according to the probability.