| In Geographic Information System discipline,geospatial knowledge is the valuable information structure from relation among geographic information of different cognitive level.Researches on geospatial knowledge is the core content of intelligentize in Geographic Information System and the base of spatial data mining and geospatial reasoning.Research mostly focuses on the algorithm of spatial data mining that acquires geospatial knowledge automatically from spatial database,and applying the geospatial knowledge to decision-supporting system in fields by geospatial reasoning.Currently,the integration method of mufti-source heterogeneous spatial data has matured,but geospatial knowledge lacks a unified representation framework.Semantic heterogeneity,spatial scale heterogeneity,representation method heterogeneity and storage structure heterogeneity exist in geospatial knowledge that acquired from different sources and different methods,which causes the difficulties in sharing and integration of geospatial knowledge,and hinders the development of deep research of spatial data mining.On the other side,the research of Knowledge Integration of Artificial Intelligence domain tends to research on business knowledge of organization,and rarely focus on the processing of temporal and spatial characteristics.Therefore,the research which focuses on unified representation and integration method of geospatial knowledge,eliminating the heterogeneity of geospatial knowledge and considering the temporal and spatial characteristics of geospatial knowledge in the integration,will provide theoretical supports for the accuracy and high-efficiency mining when facing the huge heterogeneity resources,also promote intelligent development of Geographic Information System and play a guiding role in the real intelligent city construction.According above,this thesis takes the multisource and heterogeneity geospatial knowledge as the main line,builds the unified representation model as the breakthrough point,proposes a unified representatiion model and a geospatial knowledge integration method which realizes by Conversion and merging among geospatial knowledge in the understanding of the connotation,classification and characteristic factor of geospatial knowledge.The research contents and achievements in this thesis are as follows:(1)Presents a Geospatial Knowledge Unified Representation Model(GeoKURM).And by comparing with some typical geospatial knowledge representation methods and analyzing the differences between them,this thesis takes geo-ontology theory as the base of GeoKURM.GeoKURM gives a complete description of the characteristic factor of geospatial knowledge by Geospatial Meta-knowledge,Geospatial Knowledge Ontology and Geospatial Rule Base.Proposes an improved semantic triple to describe semantic factor,a description of topological relationship based on 4-intersection model to correct deficiencies in the description of topological relationship in geo-ontology.(2)An integrated method of geospatial knowledge is proposed which is realized by geospatial knowledge transformation and combination.The general process of geospatial knowledge transformation under the transformation methods of geospatial knowledge is illustrated in this thesis,creating mapping relation described by topological relation in order to transfer with spatial description logic;As to the production rule which is close to Natural language,this thesis puts forward the method of automatic recognition and extract of implicit temporal and spatial relationship during temporal and spatial relationship vocabulary transformation to solve the representation method heterogeneity and storage structure heterogeneity of geospatial knowledge.Among the combination methods of geospatial knowledge,using the combination of GeoKURM with ontology mapping to solve the semantic heterogeneity of geospatial knowledge.By using 4-intersection model combined matrix,a solution of conflict resolution between geometrical morphology factor and temporal and spatial relationship factor is illustrated in order to solve the spatial scale heterogeneity of geospatial knowledge.Presents an improved consistency checking reasoner GeoPellet based on temporal and spatial characteristic factors in GeoKURM,realizes by the hybrid reasoning with the rule reasoner Jess.(3)The Geospatial Knowledge Manipulate Platform(GeoKMP)is designed and actualized with ability to construct domain knowledge base,as well as integrate heterogeneity geospatial knowledge source,and support integrated reasoning with actual data.Taking the synthetic pipeline data verification and the emergency auxiliary decision system of urban gas pipeline as an example,the geospatial knowledge of municipal underground pipeline were integrated into existed risk assessment knowledge base of gas pipeline.Besides,related spatial data base resource is integrated in order to optimize and improve the validity and accuracy of the proposed methods. |
PDF Full Text Request