Recommended Tourist Attractions Based On Sentiment Analysis

The renewal of new media technology has brought about changes in the marketing structure and mode of traditional tourism and opened up new development of tourism.Tourism portals emerge in an endless stream,and related data and information are growing exponentially,including visitor reviews,scenic interest ratings,personalized customized packages and so on.Massive data results in information overload and also leads to recommendation research.At the same time,tourists' comments are a form of expression of users' emotional views,which have the characteristics of openness and high value density.Group opinions provide a reference for other users when deciding on scenic spots or using services,and form a dynamic group collaborative environment.Most of the recommendation studies of tourist attractions focus on user personalization,aiming at finding relationships in social networks,mining related information of users and making similar hobby recommendation.First,the multiple value of tourism commentary has not been fully exploited,and second,the importance of group recommendation has been neglected.In view of the above situation,this paper proposes scenic spot recommendation based on emotional analysis,obtains the emotional tendency and emotional state of visitors' comments,excavates potential information and problems,and makes support for users' travel decisions.In order to improve the accuracy of emotional analysis of tourism reviews,aiming at the long and complex texts of tourism reviews,the existing algorithms for emotional analysis of tourism reviews seldom consider the characteristics of texts and the rules of syntactic changes,which lead to the reduction of classification accuracy,an improved algorithm is proposed.According to the syntactic rules,the text is directly classified into words,summary sentences and transitional sentences,which are extracted preliminarily,and then passed through the volume.The product CNN is used to classify,which effectively improves the classification accuracy.On the basis of guaranteeing the accuracy of emotional analysis of comments,emotional factors are designed to correct the imbalance proportion of negative comments,obtain thecorresponding emotional scores and standardize them.For some scenic spots,ignoring the impact of the comment base,the interval assignment method is used to adjust the coefficients.Finally,reasonable emotional scores of scenic spots are obtained and multi-dimensional data are mined to synthesize recommendation to assist users in decision-making.The validity and rationality of the proposed recommendation algorithm are proved by verification and comparison.