Linking Cultural Heritage Resources to Spatiotemporal Places

One fundamental research question in information integration in the Cultural Heritage domain is the identification of two individuals in different data sources as being the same individual in reality. This research question is also known as a co-reference problem, and is an essential question within the Linked Open Data Initiative. Places are often used in Cultural Heritage applications to link data sets, but this linkage poses semantic and practical questions. Looking at the example of a co- reference statement of Rome in the Latin Source of Tacitus from the 1st century AD and contemporary Rome, we want to investigate the identity of places, the use of place name gazetteers for co- referencing places and the spatiotemporal relations of co-referenced places. The goal of this project is to propose a theoretical and practical approach to the co-reference problem for places in Cultural Heritage applications that goes beyond the state-of-the-art. The methodological approach is based on a specific model of co-reference knowledge making the epistemological distinction of: a model of a common reality; a model of an agents opinion about reality; a model of agents opinions whether they talk about the same object or not. To model the underlying semantics of cultural heritage information and geospatial information, the approach will use CIDOC CRM and ISO 19100 standards for Geographic Information. The hypothesis for this project builds on CRMgeo, an extension to the CIDOC CRM for geospatial information, which defines identity criteria for places in Cultural Heritage. The work plan contains three work packages. The first one builds a methodology for co-referencing places while the second work package will specify and develop components of an open source tool chain with standardized interfaces to link places for Linked Open Data. The tool chain will use KARMA, a tool developed at the University of Southern California and Arches, an open source geospatial software system for cultural heritage inventory and management. The third work package which corresponds to the return phase applies the approach to link Eastern Alps mining history resources to Linked Open Data. The first two work packages will be hosted at the Information Sciences Institute of the University of Southern California as it offers the expertise as well as the semantic software infrastructure that are indispensable for the projects goals, combining the academic fields of semantic web research, geospatial information and cultural heritage in a unique way. The return phase of the project will be conducted at the Surveying and Geoinformation Unit of the University of Innsbruck which forms part of the Research Centre HiMAT that investigates the history of mining activities in the Eastern Alps. Work Package three directly applies the gained know-how and the semantic software infrastructure to the Austrian Research Centre HiMAT.

One fundamental research question in information integration in the Cultural Heritage domain is the identification of two individuals in different data sources as being the same individual in reality. This research question is also known as a co-reference problem, and is an essential question within the Linked Open Data Initiative, which tries to link data sets from different sources to each other and thus increase the knowledge. Places are often used in Cultural Heritage applications to link data sets, but the simple relation same as is in most of the cases not sufficient to represent a complex spatiotemporal reality, in particular when integrating several data sources including geographic gazetteers. The result of this project is a conceptual model and a practical approach to co-reference cultural heritage places. The conceptual model first introduces identity criteria for places in cultural heritage. They are the prerequisite for the second step to define place relations of identity (same as), being part (part of), topology (e.g. is contained in) and approximation (approximates). Places as well as relations do have a temporal extent. The practical approach applied this conceptual model to place information in cultural heritage sources and Geographic Gazetteers and was tested in the first phase with birth places of artists. An open source tool chain was developed and applied to create identity, partitive, topology and approximation relations of artists birthplaces in different sources to entries in a geographic gazetteer. The interpretation of these relations allows statements if two artists in different sources can be the same one in reality based on their place and time of birth. The goal of the second phase at the Surveying and Geoinformation Unit of the University of Innsbruck was to broaden the methodology to sources with coordinate information. An enhanced conceptual model was applied to discovery sites of features related to prehistoric mining activity. Specific structures, that may have been created during prehistoric mining activities, were extracted from geological maps, archaeological literature, objects from museum exhibitions and a high resolution digital elevation model. A co reference process using semantic and geometric criteria was executed to identify structures that are documented in several sources. Integrating the information based on the co reference lead to an increase in knowledge about the structures and their exact location. One product of this integration was a list of mines with their names and coordinates that we provided to the emergency services of the Province of Tirol. In case of an accident the names and exact locations of mines are essential to lead the rescue forces to the site of the accident. This is one example for the non-academic application of the research results of this project.