Study Of Geospatial Data Provenance Modelling And Semantic Computation

Geospatial information,as an important category of information resources,has long been used in a wide range of fields such as economic development,public health,urban planning,precision agriculture,and ecological analysis.With the development of Spatial Data Infrastructures(SDI),these geospatial resources are exposed to the public in Web environment.In the SDI,geospatial data is widely and frequently distributed and processed in unpredictable ways,research on geospatial data provenance becomes a fundamental issue.Especially in recent years,the emergence of crowdsourcing technologies has changed the production and consumption patterns of geographic information,creating enormous quantities of authoritative and volunteered geographic information(VGI)on the Web.Provenance becomes a major concern in consuming these resources varied in quality.In such context,how to express and employ geospatial data provenance in an interoperable way,and how to explore the hidden knowledge in the large repository of provenance information,are the keys to the in-depth use of professional and volunteered geographic information.Geospatial data provenance,also called lineage in Geoinformatics domain,records data sources and process steps of geospatial data products.As a crucial element of data quality,it helps to trace the sources of data products in an open Web-enabled environment and improve the trust in and reproducibility of scientific results.However,research on geospatial data provenance modelling and analysis remains challenging.First,in the aspect of the model interoperability,there is no unified and scalable model for exchanging geospatial data provenance on the Web.For instance,ISO 19115"Geographic Information-Metadata" standard shows poor interoperability in the Web environment,while W3C PROV lacks domain-specific semantics.Second,from the perspective of computation,traditional geographic information science neglects the conceptual recognition of provenance,and current geospatial computing frameworks are difficult to support provenance operations.Last,in terms of analysis and application,there is still no theory or methods about geospatial data provenance analysis,which is extremely important for the use of VGI information.These are three major reserch issues lying ahead in the field of geospatial data provenance.In order to solve the above problems,this thesis aims to establish an interoperable geospatial data provenance model using Semantic Web technologies,addressing key issues involved including spatio-temoporal provenance,granularity of provenance,meta-provenance,and identification of mutable geospatial resources.Then the paper tries to extend the existing framework of core concepts of spatial information,by introducing the concept and core operations of provenance.After that,the author attempts to summarize the basic theory and methods of geospatial data provenance analysis,based on which quality assessment of VGI is conducted.The contributions of this thesis can be summarized as follows:(1)A semantic mapping approach is proposed for geospatial data provenance modelling.Based on several use cases,the paper analyzes two popular provenance models adopted in the Geoscience community:W3C's PROV and ISO's lineage model.It discusses key considerations when moving geospatial data provenance into the Web.A mapping framework is established and mapping rules are given,in order to support cross-domain query and reasoning of geospatial data provenance.(2)A provenance-aware framework for spatial conputing is established that supports core operations of provenance.Based on the idea of core concepts of spatial information,the paper introduces the provenance concept and defines core operations on provenance.Then a semantic API is implemented to support spatial and provenance computation.(3)A theory on spatio-temporal data provenance analysis is presented.In the existing literature,there are few studies on geospatial data provenance analysis.A list analysis method of geospatial data provenance is given in this paper,two of which are demonstrated for the assessment of OpenStreetMap data quality.The work in this paper provides new ideas for quality assessment in the collaborative Web environment where volunteered geographic information is widely used in combination with professional geographic information.