Vague Modeling Method Of Place Names Based On Web Pages

As one of the most commonly used social and public information, Geographical names are the essential foundation for state administration, economic construction, domestic and international exchanges of information, and have rich applications in mobile location services, network, electronic maps, intelligent transportation, public safety and other areas. At present, there are many geographical name databases in China with different levels and scales, such as national, provincial, municipal and county level. However, due to the large number of China’s place names, rapid renewal of place names, geographical names database information is incomplete, especially the the widespread missing information of spatial locations. With Internet technology system such as Web2.0appeared, the network has become a new type of important geographic information sources except of GPS, remote sensing image. Acquisition and interpretation of spatial location information on the Internet has become important content of the geographic information science.In this paper, the CRF Model and Rule Model realize the acquisition of spatial information in the web page text. On the basis of the spatial semantic disambiguation of geographical names, the paper discusses how to develop the vague modeling method based on density analysis and the vague modeling method of semantic analytical model of spatial relationship. The main research results are as follows:(1) Acquisition of spatial location information of geographic names:With the information in the network as data sources, on the basis of place name recognition and spatial relations extracted using a Web crawler technology and theme of the sensitivity of Web access to technology, and then draw on natural language processing technology, the place names in page text entities, places coordinates information spatial relationship information extracted. Fuzzy modeling provides a data base for the further place name disambiguation and place names space.(2) Spatial semantic disambiguation of place names based on their administrative subordination tree:according to the names of the places in this paper have certain space relevance and the names of the places which are included in the geographical name data bank have the administrative levels of subordinate relationship, so based on recognition of geographic names in the web page of text, we can obtain ambiguity places at close range of context geographic names. Geographic names match based on database, construction and ambiguity geographic names associated administrative level tree structure, and according to the tree structure of the robustness to determine the location of ambiguity geographic names. Experiment indicates, in web pages which are sensitively, the method is feasible.(3) Vague modeling of place names based on density analysis:Firstly, this paper extracts place names from the Web pages based on CRF model of Natural Language Processing. Secondly, a density analysis of these place names with the spatial correlation is made. Finally, spatial vague areas are constructed with the density threshold of the density point. The experiment results show that this method can achieve the vague modeling of place names, but with a limit of the quality of place names from Web pages. However, this method has a better complementarity with the modeling method based on spatial relations, both of them can improve the accuracy and reliability of the vague modeling of place names..