Research On Travel Recommendation Method Based On User Sentiment Portraits

The tourism economy will become a new driving force for the economic development of Shanxi and many other provinces during " The Fourteenth Five-Year Plan" period.Cultural and characteristic tourism will also become the mainstream and the upgrading of tourism consumption will become more obvious during this period.It has become a research hotspot for tourism websites and scenic spots to better recommend local tourism resources for tourists with different needs.This article mainly uses sentiment analysis methods to analyze tourists' ratings and comments on different scenic spots.The review mining algorithm and collaborative filtering algorithm are researched and further improved.And a new method of recommending scenic spots based on user profile is proposed,and the algorithm is proved to be highly accurate through experiments.The main content and contributions of this paper:(1)This paper proposes a comment mining algorithm based on user profile and uses an algorithm to automatically discover part of speech for the extracted comments.Perform text preprocessing on the mined comment data and extract opinions from it.Then use the ASUM sentiment-topic mixed model to output the user sentiment-topic distribution matrix.(2)A collaborative filtering recommendation algorithm based on user profile is proposed for travel recommendation.The similarity calculation formula is improved to multiply the cosine similarity and the Jaccard coefficient.Use a hierarchical clustering algorithm to classify users,and then calculate the overall similarity between users and scenic spots.Finally,this article will predict the user's preference score for unknown scenic spots and generate recommendations based on the score.(3)This paper uses a Python-based crawler model to crawl some datasets,which are the tourism data of Shanxi Province and the data of users' comments and ratings on scenic spots.Then design a comparison experiment,which compares the algorithm proposed in this paper with the recommendation algorithm based on the traditional LDA model and the collaborative filtering recommendation algorithm based on user ratings.The results show that the algorithm proposed in this paper has high accuracy for recommending scenic spots.