Research On Attraction Recommendation Algorithm Based On Social Trust And Tags

With the rapid development of the Internet,people can select their favorite travel products and formulate corresponding travel plans on various online travel platforms.According to iResearch's test data,the transaction volume of China's online travel market reached 738.41 billion yuan in 2017,an increase of 24.3% over the same period of last year.In 2017,the online travel OTA market in China was 40 billion yuan,a 34% increase over 2016,the proportion of future online travel will further increase.However,due to the obvious development of Internet technology,the problem of information overload makes it difficult for users to quickly retrieve the tourism information they need from a vast amount of Internet information.At present,the online e-commerce platform for online travel services basically stays on simple information search.The service items are single and the introduction of tourist routes and scenic spots are almost always fixed.Many traditional methods such as texts and pictures are used to describe the concepts in a rough way,which fails to satisfy user's personalized needs.Moreover,when these platforms provide users with tourism product search services,the search results are ranked by the overall score and ignore the user's own interest,so it is difficult for customers to obtain personalized services.Aiming at the existing interest point recommendation algorithm,we find the user's implicit trust and trust transfer problem are neglected when dealing with users' relationships.And it is tough to make an accurate recommendation for users at a new city due to the lack of user history records.Aiming at these problem,this paper proposes an a personalized attraction recommendation methods integrating user trust and label preferences.First,trust is introduced when the user similarity is poor to satisfy the recommendation quality.And we further mine the users' implicit trust relationships to solve the dilemma when the direct trust is difficult to obtain.The results shows data sparsity and cold start problems are effectively alleviated.Next,we decompose the users' interests into lang-term preferences for different attraction tags by extending the relationships between attractions,tags and users,which effectively alleviating the lack of user history tour records and new city problem.Experiments were performed on the data collected from the Flickr website.The results show that the proposed hybrid recommendation algorithm effectively improves the recommendation accuracy and mitigates cold start and new city problems to some extent.