A Collaborative Filtering Recommendation Model Based on HMM

Considering that browsing path, access time, visiting times and etc are the important factors to give influence on accuracy of recommendation, a dynamic collaboration filtering recommendation method based on Hidden Markov Model (HMM) is proposed. First, it simulates a customer's behaviors while the customer is browsing web pages, and sets up the nearest-neighbor set according to his behaviors. Because of the used data is not a customer's scores, but a customer's browsing paths, the problems such as data sparseness and initial scoring are solved. When HMM is used to replace the similarity model to measure a customer's similarity, the accuracy of nearest-neighbor recommendation is improved greatly. And it settled the on-time recommendation problem and the extensible data space problem. Then the concept of preference degree is set up. As the customer's preference degree of is introduced, the item recommended to a target customer is more suitable. Finally, the preference degree is applied to establish the prediction model of dynamic collaboration filtering recommendation. An experiment shows the excellent performance of this model.