Research On Intelligent Tourism Recommendation Based On Guilin Scenic Spot Data

With the rapid development of the economy,people's living standards have also increased,prompting the rapid development of the tourism industry.Tourism travel accounts for a large proportion of the tourism industry chain,so it is especially important to provide intelligent travel.Personalized travel recommendations are increasingly becoming an important topic in the development of attractions.However,the user's interest in the attraction and the interest points generated by the catering accommodation are quite different.It is difficult to know the next determined target of the tourist,and it is greatly affected by the irresistible factors such as weather,traffic and traffic.At present,there are many tourist service data and various types of scenic spots.Facing with such huge tourism resources information,it takes a lot of time for tourists to inquire about tourist information,which will only bring great inconvenience to tourists.Therefore,in order to solve the above problem that the user lacks the directionality in the conventional retrieval system,it is necessary to construct an intelligent travel recommendation system.The rapid development of “Internet+” tourism has given birth to a large number of information such as reviews,ratings and pictures of tourist attractions,which have become an important reference factor for users to choose tourism products.The traditional recommendation only considers the single case of user ratings,and the traditional recommendation model has problems such as low recommendation rate,low efficiency and insecurity,which ultimately leads to the recommendation results are often unsatisfactory.In order to solve these problems,this paper proposes the following methods:(1)Due to the large number of users and tourist attractions,the relationship of user and the attraction is a sparse matrix.For the sparse problem of tourism data and the low accuracy of the traditional recommendation algorithm,a theme-based factor decomposition machine recommendation method is proposed.By comparing the score-based collaborative filtering recommendation with the topic-based collaborative filtering,the theme-based factor machine model has higher accuracy and recall rate;(2)Considering the low recommendation rate and the single rating problems in traditional recommendation model,and the existing user review information and the tourist scoring,this paper based on the traditional collaborative filtering model proposes a mixed model which integrate the user's sentiment scoring with attraction popularity and attraction theme scoring;(3)Based on the problems of low efficiency and insecurity in the traditional collaborative filtering model,a blind signature algorithm is proposed by integrating KMeans clustering algorithm with RSA on the collaborative filtering model.K-Means clustering algorithm can improve the low efficiency of traditional recommendation system.Blind signature algorithm is used to encrypt the recommendation result,which protects user privacy to a certain extent.