Research On Individualized Tourism Recommendation System Based On Diversity

Till now, the recommended system has developed enough to solved numerous problems of overloading information for many travelers, especially with coming of big data and mobile Internet era.The personalized traveling recommendation system, based on characters like historical records of tourists and demographic statistics, establish a user preference model so as to recommend relevant information for travelers. Despite of these achievements, this kind of system has still faced many challenges, one is that the current system cannot accurately and comprehensively describe travelers’ preference; another one is that users who get used to the traditional recommendation system that pursues high accuracy are not able to completely adapt to the new one.Aiming at the problems above, this paper focuses on researching the users modeling of personalized traveling recommendation system based on the combination of users’ attributes and domain knowledge, as well as researching methods to improve the diversity and novelty of recommendation by extending users’neighbor selection. The research is carried out as follows:1.Research on existing users’ modeling in personalized traveling recommendation system and put forward extend vector mode based on vector space mode. This new modeling method explores the users’ demographic statistics and intersection record of users and items, and undertakes filtration of other redundant information, in order to accurately describe travelers’ preference and improve cold start issue of the system.2. Explore on current improvements on diversity and novelty in recommendation system and come up with a new recommendation algorithm based on nearest neighbor selection. This algorithm divides users into intra-cluster and inter-cluster and calculates the popularity of the whole items in inter-cluster. Meanwhile, it takes user diversity in intra-cluster into consideration, and proves the effectiveness of this algorithm by experiencing on real data set.3. Integrate with the modeling of various and multidimensional users’ preference to build a personalized traveling recommendation system, giving examples of tourist attractions in Shaanxi Province. It provides effective recommendation service for travelers by data gaining, data analyzing and processing, user modeling and recommendation achieving. At last, the author, based on the algorithms in this paper, designs a new recommendation system for tourist attractions. It integrates users’ basic information with actual data and provides characterized recommendation for users. Meanwhile, it solves the clod start problem with the application of tag cloud.The algorithm and modeling put forward in this paper is not only limited to accuracy achieving. Furthermore, it considers, when offering recommendation to travelers, various elements such as specific knowledge of this field, users’ features and so on. As a result, such system which applies a diversity improved algorithm is able to give a novel and content recommendation to travelers. And it is of certain value and can provide a reference for other papers which aims at solving similar issues.