Point-of-interest Recommendation Model Based On Geo-tagging And Sentiment Analysis Of User Reviews

In recent years,the product recommendation model integrates more user personal information,which effectively enhances the cohesion between users and products,and improves the accuracy of recommendation.Personalized recommendation algorithm based on user information has become a key research direction.For different types of user information,the recommendation algorithm also presents diversity,and the recommendation method based on regional division and emotional analysis gradually reflects an increasingly important position.Different geographic areas affect the changes in users 'interests.At the same time,the time factor has a certain effect on the change of the polarity of comment emotion,which in turn affects the user's choice.Some studies have not focused on these issues.In response to the above problems,the main works of this paper include the following four aspects:1.This thesis proposes a Point-of-Interest Recommendation Model Based on Geo-Tagging and Sentiment Analysis of User Reviews(GTSA).Realize the user's local and foreign recommendation.2.Through the expansion of the content description text information of the points of interest,the measurement of the similarity between the points of interest is realized,thereby constructing a content-based recommendation model.3.A set of methods for sentiment mining on unlabeled comment data is proposed to obtain the sentiment polarity preference of point-of-interest reviews,and the time window and the sentiment value of user reviews are combined to analyze the sentiment polarity of the overall review of interest points.4.In the recommendation list sorting stage,two sorting methods of interest points were constructed,a sorting method based on local users and a sorting method based on foreign users.In the local recommendation,we combine the emotional trend of the overall comment with the similar interest points based on the user's historical records,establish a new scoring model,get the recommendation score of interest points,and then get the localrecommendation that conforms to the user's personal interest.In the field recommendation,we combine the emotional polarity of interest points with the number of comments,propose the measurement method of interest point temperature,and mix it with the scoring method of local recommendation to achieve the recommendation of users' field interest points.Experiments were performed on the data sets Yelp and Foursquare.The results show that the GTSA model presented in this paper has a certain improvement in the accuracy of recommendations.