Recommendation System Research Of Users’ Browsing Path Combining Ant Algorithm

In the age of knowledge explosion, the network information increases rapidly. Simple search engine has failed to meet the need of user getting useful data from massive information on the internet, which lead to lower utilization of information. In order to solve this problem, researchers put forward recommendation system, which can effectively solve the problem mentioned above. Recommendation system would provide a recommending list for each user through the analysis of the history of the users’ behavior that indicate users’preferences, habits and establish corresponding recommendation algorithm, which could help users to find the information they are interested in quickly.This dissertation first introduces the major recommendation algorithms of the recommendation system and the advantages and disadvantages of these recommendation algorithms In order to improve the quality of recommendations, a new recommendation approach is proposed combining ant algorithm and the Collaborative Filtering in this dissertation. According to the principle of ant foraging, the user is considered as "ant", the target commodity as "food", the next commodity to be browsed is predicted by the pheromones between the ants’ communication. A experiment is designed to test the prediction accuracy and classifiction accuracy of the algorithm.The accuracy is not the only criterion to test a recommendation system’s perfomance, this dissertation also introduces the recommendation diversification into the recommendation system to improve the users’ friendliness and make the recommendation system more personalized. The benchmark Movielens is applied in the experiment, the results shows that this method can efficiently alleviate the dataset sparsity problem and provide better recommendation results and recommendation diversification.