Research On The Recommendation Algorithm Of Interest Points That Integrates Friend Ratings And Reviews

In order to make the recommendation of interest points more and more intelligent and increase people's satisfaction with the recommendation results,more and more scholars have studied the recommendation algorithm of interest points.The existing interest point recommendation algorithm,because the factors affecting the user's score are not considered comprehensively,leads to a bias in the analysis of the user's preference for points of interest.Furthermore,most scholars do not analyze the user's evaluation.Since the user's evaluation content can reflect the user's favorite point of interest feature,ignoring the user's evaluation content is not conducive to analyzing the user's favorite interest characteristics,and It will affect the result of recommendation.Aiming at this situation,this thesis conducts sentiment analysis on the evaluation content of friends,analyzes the characteristics of interest points that users like,and proposes a score prediction algorithm based on friend evaluation.According to the type of influencing factors,the user score is modeled.Through the fusion of score and evaluation,an interest point recommendation algorithm that integrates friend score and evaluation is proposed to recommend users with unvisited and possibly higher intentions.The specific tasks are as follows:First,a score prediction algorithm based on friend evaluation is proposed.Due to the randomness of the user's check-in,this paper filters the user's interested friends through the location information entropy.By comparing the set of points of interest of the interested friends with the user,the points of interest that the friend may recommend for the user within the time threshold are selected.The word segmentation technique is used to process the evaluation of friends.Using the idea of the topic model,the sentiment analysis of the evaluation content from the attribute word to the attribute face is considered.The friend's attention to the feature point of the point of interest is considered,and the evaluation content of the friend is modeled.Analyze the characteristics of the points of interest that the friend likes and calculate the predicted score of the friend's evaluation.Second,a point of interest recommendation algorithm based on friend relative score is proposed.This article analyzes the factors that affect the friends 'score,models the friends' score,and obtains the corresponding score value to reduce the impact of the score deviation on the recommended results.By Fusion parameters,the corresponding score of the score and the predicted score of the evaluation are merged to obtain the fusion score.The probability of the point of interest being recommended is calculated by the probability point recommendation probability algorithm based on the encounter idea.The product of the fusion score and the recommended probability is taken as the relative score,and the relative score is sorted according to the size of the relative score.Finally a interest point recommendation algorithm for fusion score and evaluation is formed.In the accuracy and recall rate,the algorithm is compared with other algorithms based on scoring and review.The experimental results show that the proposed algorithm performs better.Therefore,the interest point recommendation algorithm based on friend rating and review proposed in this paper can better analyze the user's favorite points of interest and their characteristics,and can recommend the points of interest that the user may like and have not been,and can better satisfy the user's needs of the point of interest.