Research On Path Mining And Retrieval Based On Massive Spatial-Temporal Data

The development of information technology makes people’s modern life more convenient. With the development of the Internet technology, especially the Web2.0technologies such as blogs, social networks, micro-blog, and the role of Internet users are changing. Internet users began to slowly change from consumers to producers. Different with accessing network resources edited by others, sharing become the trend. Experience shared by uses all over the world constitutes a huge knowledge base. How to use the huge information to build service for people’s daily life, is a very valuable research problem.This dissertation starts with huge images shared by Internet users, mines users’ travel paths from images, establishes a search engine for the mined paths and provides users with personalized recommendation service of route based on users locations.The main research content and innovation are as follows:1. Research background and current situation of temporal-spatial data was analyzed. The development of GPS system, especially our "Beidou" system, was summarized first. Researches about user-generated geo-tagged multimedia were surveyed. Problems about the management of massive temporal-spatial data were summarized;2. Proposed a kind of fast and efficient user route mining algorithm from massive images. The algorithm takes advantages of metadata such as time and space information of images, according to the exact time and location generated when the image was shooting, and restores the location and location sequence of the person. Because tourists usually only share a subset of their photos, paths mined from photos cannot repeat the true path of tourist. Search-based path enrichment algorithm was proposed to combine different users path and get better paths;3. Proposed a path index algorithm, a fast route similarity calculation method and a random walk with restart path ranking algorithm to construct a path retrieval system. Paths are new special resources. In order to provide location based services (LBS), a spatial index algorithm of paths was proposed first.The algorithm takes into account not only the route location information, but also the data structure of paths. This dissertation puts forward to directed sparse chamfer distance, which can be efficiently calculated in linear time, to calculate the similarity between paths. This two algorithms make the real path retrieval system possible;4. Proposed a personalized path recommendation algorithm based on users’ history location information.on-site travel guide-a mobile app which can be deployed on a mobile phone to record users’trajectory and provide instant suggestions to help users to choose a direction to follow;5. Proposed a novel interactive trip planning system to enable users to plan their trips in a more efficient way. Users can visually view paths on a map and photos along the path. Moreover, users are allowed to interactively adjust the suggested plans by adding or removing destinations to get more customized travel routes from the system.The dissertation ends with a summary of my work and an outlook in this research area.