Research On Personalized Route Mining Based On Users’ Preferences

The development of location acquiring technology results in the generation of a large amount of spatial-temporal data.In recent years, the research on these data gets more and more attention. Photograph sharing becomes one of the most popular Web services. Many websites of photographs sharing provide photos with labels, which makes photograph organization more convenient. People record the things attracting them by taking photos, which can form the photos with location labels. A photo can provide rich information including a users’ location and time. The thesis mines users’preferences based on users’ routes formed by photos. And, the personalized route is mined, taking most people’s routes.To describe users’ preferences, the thesis imports TF-IDF algorithm. On this basis, an evaluating function is defined based on a user’ interest and the frequency of a route to mine the route meeting the user’s preference. Moreover, the shortest path between two adjacent interesting points is mined based on the traffic network to recommend to users. The main contributions of this thesis are summarized as follows:First, TF-IDF algorithm is imported to extract users’preferences. Interest points are replaced by interest categories to describe routes. And, the contribution degree of an interest category to a user is regarded as the users’ interest to the category.Second, for the storage of routes, an indexing structure of route segments is proposed, which reduces the cost of storage space and improves the mining efficiency of routes.Third, for the route choice, the thesis proposes an evaluating function based on users’ interests, which considers both users’preferences and frequencies of routes. Based on this evaluating function, the route meeting a user’s preference can be mined.Finally, for the route recommendation with the limit of time, a dynamic recursion algorithm is presented. On the basis of pruning candidate routes, the route is chosen dynamically according to the evaluating function. And the thesis employs the improved A* algorithm to mine the shortest path between two adjacent interest points in one route to acquire the best route.Theoretical analysis and experiments validate the personalized route mining method based on users’ preferences proposed by the thesis is effective.