The Research Of Personal Recommendation Algorithm Based On User Comment

The traditional recommendation algorithm is based on the user rating or the attributes of the items and other information to do personalized recommendation,but the statement data of items often contains more valuable information,it is more useful to dig the characteristics of users and items,and help us to establish complete portraits of users and items,so as to provide more accurate personalized recommendation for users.In this paper,the statement data of the items are introduced into the traditional recommendation algorithm model,combining it with other information(the information of the items,the demographic information of the users,and the user's historic behavior data)to calculate the similarity between users and items,and then achieve recommendation algorithm based on user comments.The weight combination of the multi-dimensional information describing the similarity between users and objects is solved by genetic algorithm.It is expected that the traditional collaborative filtering algorithm can improve the predictive accuracy of the recommendation algorithm,while solving the problems of user cold start and item cold start of the traditional algorithm.Finally,this paper use the Multi-model fusion algorithm to improve the prediction accuracy of the model,the experimental results show that the GDBT algorithm can improve the prediction accuracy of the model.