Research On Multi-source Global Toponymic Data Fusion And Updating Methods

As an indispensable kind of information in geography and social public,toponymy plays an important role in national and social management,economic development,cultural construction and national defense diplomacy.With the development of computer technology and the popularization of mobile Internet,the way of collecting and serving.toponymic data has changed greatly.At present different countries,institutions or enterprises have established various types of global toponymic databases,and most of them provide query and sharing services on the Internet,such as GeoNames,OpenStreetMap,Geonet Server Names and so on.However,these databases are quite different in coverage,form,language type,content and so on.At the same time,they have remarkable complementary advantages.Therefore,it has become one of the basic problems to be solved urgently in the current geomatics data mining and utilization on how to use these open global toponymic data resources to build a more complete and rich global toponymic database with coverage and data content.In view of this,this paper proposes a multi-source global toponymic data fusion and updating method by constructing a toponymic similarity model considering semantic features and a multilingual feature-oriented toponymic index effectively,which enhances the global toponymic database in ways of integrity,accuracy,reliability,actuality.The concrete research contents and achievements mainly include the following aspects:(1)place name similarity calculation model considering semantic featuresTaking GeoNames,OpenStreetMap(Osm),GEONet Names Server(GNS),DIVA-GIS,Geographic Names Information System(GNIS)as data sources,the acquisition and preprocessing methods of multi-source global geographical Names data are studied.By improving Edit Distance Algorithm and Greedy String Tiling,this paper constructs a model for calculating the similarity of place names.A model for calculating the spatial similarity of place names is constructed by considering the spatial distance,the located administrative territorial entity and the type of place names.On this basis,a model for calculating the similarity of geographical names based on the combination of geographical names and spatial features is constructed,which effectively solves the problem of judging the consistency of different toponymic data and realizes the fusion of multi-source global toponymic databases.(2)multilingual feature-oriented gazetteer index building methodIn view of the features of the phonetic and ideographic characters of different languages,the language-detection language detection database is used for the recognition of place names.The linguistic features of English place names,such as the total number of letters,the initial number of letters,the total number of words and the first letter coding of words,are analyzed and the index organization method based on multi-dimension feature statistic vector are studied to solve the problem of index establishment of phonetic languages.The language features of Chinese place names,such as the same characters,the number of characters and the position of characters,etc,are analyzed,which led to the studies of the organization of Chinese place name index based on single Chinese character.It solves the problem of index establishment of Ideographic toponymic language.(3)prototype system development and Experimental Verification AnalysisA multilingual global toponymic database(more than 25.21 million in total)has been constructed based on Geonames,OPENSTREETMAP(OSM),GEONet Names Server(GNS),DIVA-GIS,Geographic Names Information System(GNIS),and Global Administrative Areas(GADM),the website of global administrative territorial entity,and a book of gazetteer in 21st century.On top of that,a prototype of global toponymic retrieval system has been developed to realize the functions of multilingual global toponymic information retrieval,map and statistical analysis,and shared interface.By using the recall rate,precision rate and efficiency,the model of semantic similarity and the index organization method are tested and analyzed.