Study On The Semantic Association Model Of Geospatial Knowledge Based On Linked Data

Through the existing land cover data and information service platform,users can obtain intuitive spatial distribution and temporal and spatial changes and other information,but for its distribution characteristics,geographical distribution,change factor,development trends and other professional knowledge used to assist decision-making analysis,it is difficult to obtain.In order to meet the needs of comprehensive geoscience analysis,disaster prevention and mitigation,and government decision-making,it is necessary to associate multi-source knowledge of different disciplines,specialties and applications.The main bottleneck of realizing multi-source data association and accurately discovering potential association is the semantic heterogeneity of geospatial data.How to build association model is the key technical issues and how to associate unstructured knowledge of other fields with geospatial knowledge together.It is found that the traditional multi-dimensional correlation model based on spatial features,time-series correlation model based on time features and multi-scale correlation model based on scale features all have imperfect relationship expression,so it is difficult to achieve more extensive and deeper association.In view of the above problems,we need to start from the semantic description of geographical entities,mining out the potential relationship between geographical entities,and establish a geospatial semantic association model.On this basis,the paper analyzes the shortcomings of three most common geospatial semantic association models based on keyword matching,RDF and ontology,and proposes a linked data based geospatial semantic association model combining RDF and ontology.The main research work and innovation of this paper are as follows.(1)A knowledge object extraction method based on self defined dictionary is proposed.Through the model training,we can get the custom word list,and add the word list of surveying and mapping and geographic science to divide the word by the rapid dictionary method,which improves the accuracy and efficiency of the word segmentation.(2)A geospatial semantic association model based on linked data is proposed.Firstly,ontology is constructed automatically based on text similarity calculation.By training the spatial vector model,we extract various subject relations,spatial relations and temporal relations between knowledge items,and build the GUK ontology through classes and relations.Then,the ontology and RDF are combined.Through the defined GUK ontology,the corresponding RDF is generated,and the knowledge itemsin the database are mapped with the classes in the associated data to complete the transformation from the traditional data to the associated data,and then the semantic search is realized.The experimental results show that the linked data technology is applied to build semantic association model,mining the potential association relationship between different subject data,and improving the search efficiency.(3)The urbanization knowledge prototype system is developed.Combining the published GUK association data with the surface coverage data,while obtaining the information of the spatial distribution and temporal and spatial changes of the surface coverage,part of the urbanization professional domain knowledge can be obtained at the same time to assist the user in decision-making analysis.