User Characteristics Classification And Restaurant Location Service Recommended Method Of Collaborative Filtering

With the recommendation system pursuit more of the of individuality,using tradetional collaborative filtering recommendation as the basis of the user preferences can not fully consider the characteristics of the user,the measurement of the user’s interest is not accurate,and can not be completely to meet the user’s personalized service needs.Since ancient times,China’s food culture is broad and profound,and has a large geographical differences,each person’s eating tastes will be born or the impact of living environment,the choice of the restaurant is not limited to their own spending level,but the taste of the restaurant,the environment,the level of service and so put forward a higher level of requirements.Based on the characteristics of the user,this paper constructs the user ’s eating interest model by using the bayesian classifier,and cooperatively filters the restaurants in the surrounding area according to the user’s taste.In this paper,naive bayesian statistical methodology and collaborative research based on items are used as the recommended algorithm basis,and the user recommendation model is generated on the training set.Finally,the test set is validated and evaluated.The experimental results show that compared with the traditional object-based collaborative filtering algorithm,the proposed method has improved and improved the proposed accuracy,cold start and algorithm efficiency.This paper mainly from the following aspects of the research work:(1)A comprehensive analysis of the application of personalized recommendation of the status quo and its existing problems,combined with the location of food and beverage service recommendations,clearly put forward the contents of this study and objectives.(2)In depth study of the classification and principle of collaborative filtering algorithm,comparative analysis of the existing algorithms and summarize the future research trends.(3)Bayesian decision theory is introduced into the recommendation algorithm,and the mutual information and information entropy are used as the influencing factors to improve the classification of Naive Bayesian classification.(4)Combined with the public comment network comment data for experimental design.On the basis of the user’s preference model,select the restaurant score data which is in the user’s surrounding area to train the user’s taste,train the recommend-dation system of the diet to give the top Top-N restaurants as the recommended results,and push their location information to user.(5)Evaluate the accuracy of the experimental results,the accuracy of classification and the efficiency of the algorithm by using the evaluation index of the recommended system,and whether it meets the expected target and the shortcomings.