03 Sep 2012 to 03 Sep 2012
SDI development efforts are typically focused on improving spatial information discovery and access. With the proliferation of SDI initiatives the “Yet Another Portal” (YAP) pattern has emerged with each initiative providing yet another mechanism for cataloguing and enabling users to search for spatial information resources. As they often provide coarse-grained and incomplete metadata about available resources the current status of SDI is thus analogous to an antiquated library catalogue.
Current approaches suffer from several key limitations. Firstly, SDI architectures do not provide a means of referencing or searching for a specific feature (for example, the city of Sydney) without first knowing the location of information source for the feature and the form in which it is represented. SDI interfaces- such as OGC WFS or proprietary equivalents- provide data from a spatial representation perspective, but do not provide identifiers that can easily be cited or used across system boundaries. In the Linked Data environment for example, most services utilise geonames.org data. This resource does not provide stable identifiers for places nor does it provide access to feature instance metadata. The result is that the identity of a feature is scoped within a particular dataset and representation and there are inadequate mechanisms to reconcile and associate the multiple identities and representations of the same real world feature in use.
Secondly, current approaches to handling metadata at a data set level is too coarse grained for anything other than the most basic data discovery and there is a lack of feature level metadata. This has some implications for end users who, in order to assess fitness of use of information for particular application(s), need access to metadata related to data structure, semantics, quality and completeness.
Finally, features delivered in information resources through SDI, are not well integrated with information systems that deliver statistical information about those features. For example population projections for Australian cities contained in an online application are often disconnected from the spatial data sets that provide geospatial information and different representations for the cities.
This paper presents a Linked Data approach to managing and using feature level identifiers to link information systems to SDI information resources, being developed through the UNSDI Gazetteer Framework for Social Protection in Indonesia Project. This approach uses Web addresses (URIs) for identifiers (either place names or feature asset numbers such as postcodes) that can be used to access feature information using traditional SDI data access services such as WFS. These identifiers are used to explicitly reference individual features. This is analogous to a citation database, enabling explicit reference to identified real world features within a particular data set. The paper then provides an overview of how these identifiers can be conceptualised as ‘spatial bookmarks’ and used to link multiple representations of the same feature in different layers of a traditional SDI, as well as linking spatial features to statistical information held in systems beyond SDI boundaries.
The paper also describes a proposed approach to assessing feature information (datasets or instances) fitness for use by determining which data sets use the feature information as spatial reference(s). Links between datasets are able to be displayed on an ‘information network map’ and users can interrogate and query the associated databases and systems utilising identifiers from any source in the linked data network. The benefits of this approach are numerous and include improved quality of and access to information and the ability to more rapidly discover, integrate and analyse related information. The authors propose that through use of the mechanisms described in this paper, SDI can realise its potential as a spatial enablement mechanism that supports linking of multiple information systems, rather than remaining as a platform to deliver collections of spatial representations.
Download the Presentation [PowerPoint Presentation - 9.35 MB]