Research And Implementation Of Personalized Spot Recommendation System Based On Topic Model

With the rapid development of Internet technology,obtaining interesting tourist attractions through the Internet is becoming an important way for people to choice their travel destination.In the age of information explosion,how to rapidly filter valuable information from the Internet has become an urgent problem to be solved.At present,the traditional collaborative filtering algorithm still has little effect on solving the problem of sparseness and recommendation accuracy of the tourism attractions recommendation system data set.This paper fully considers the impact of user's review information on the accuracy of tourist attractions recommendation.Based on the sentiment analysis,the LDA model is used to deeply study the texts of scenic spot reviews,user's emotional preferences,and the generation of attractions recommendation results.The research work covers the following three aspects:(1)Analyzing and comparing current key technologies and research status related to data mining and natural language processing,and studying the currently available recommendation technologies for tourist attraction recommendation systems.Therefore,the problem of how to construct the user interest preference model based on the user review text in the tourist recommendation system based on the topic model is introduced.(2)Proposing a personalized tourist attraction recommendation method(TMSA-CF)that integrates emotional factors in the LDA model.Combining the syntactic analysis tree to complete the syntactic dependency analysis and extract the feature words on the processed tourist commentary text data set.According to the polarity of affective words and the degree of adverbs,the scores of the corresponding feature words are predicted,the feature words are classified as potential topics based on the LDA topic model,and the scores on the potential topics are predicted according to the scores of the feature words,and the traditional collaborative filtering is adopted.(3)Through comparative experiments on real data sets,it is found that the recommendation method based on the LDA topic model(LDA-CF)is more accurate than collaborative filtering recommendation method(CF),and the TMSA-CF studied in this paper is more accurate than the LDA-CF,and it also compensates for some of the defects due to sparse data.Finally,a personalized tourist attraction recommendation system for Hainan tourism resources was implemented,and the system was functionally tested.