Semantic Change Detection Of Geographic Information Based On Web Pages

This thesis is supported by the National Science and Technology of Surveying and Mapping Project "Web Geographic Information Searching and Extraction". It explored the webpage based geographic information semantic change detection method, using analyzing the Chinese expression characteristics of webpage based geographic information changes and extracting the information of place names, times and attributes. The experiment results were finally discussed and analyzed, which got from the prototype system. The main research methods and results included the following three aspects:(1) Knowledge bases for representation of geographical information changes. By analyzing and summing up the knowledge of basic geographic information change detection, several vocabulary bases were completed, including geographic information element feature words, element updating words, stop websites, stop words, spatial relationship words and etc. Relationships among the bases were also defined. On this basis, protege and OWL API were used to develop and manage the knowledge bases.(2) Extraction of geographical information changes of web pages. The two crawlers, one for Google Search API and one for general themes could solve the problems of span mining and depth mining of related web pages efficiently. And the webpage credibility calculation model could easily support the evidences to distinguish the false web pages. In addition, this paper stated a geographic information change detection method based on knowledge base, using time extraction and place recognition.(3) Prototype system and case study. By using rich client technology, the functions of the prototype system were designed with the management of knowledge bases, the acquirement of changed information and the spatial analysis. And the feasibility of the method is verified by taking Nanjing district as an example.The research indicates that the knowledge bases have some limitations, one is the complexity of the knowledge bases which are caused by so many kinds of features and relationships are included in, the other is the deficiency of the knowledge bases which are caused by the manual construction. Then, different kind of geographic information element has the different occurrence frequency in web pages. And the kinds of transportation and water are proved to have the highest accuracy of processing. Besides, most attributes of geographic information in web pages are qualitatively described. And as the attributes are with strong fuzziness and uncertainty, other data resources are needed to verify the results.