An Approach to Building User Rating Behavior Model Based on Bayesian Network

In recent years, various network applications have accumulated a massive amount of user rating data, analyzing user rating behavior and establishing user rating behavior models, which cannot do without the support of these data. To further explore the relationship between user rating data and user rating behavior, a user rating behavior model construction method based on Bayesian networks is proposed. Firstly, we provide a Bayesian estimation algorithm for the parameters of the user rating behavior model; secondly, starting from the initial borderless model, the K2 algorithm is used to search for the initial borderless model and generate candidate models; finally, considering the degree of fit between the rating data and the model, the Cooper Herskovits (CH) scoring strategy is used as the measurement standard for model and data fit, and the CH scoring optimal model is selected as the end user rating behavior model. The experimental results indicate that the method proposed in the paper is efficient in constructing a user rating behavior model.