Formal Languages for Expressing Data Consistency Rules and Implications for Reporting of Quality Metadata

Over the past thirty years, in excess of 36 billion dollars has been invested in the collection of spatial information across the public sector in Europe (PIRA, 2000). There is a clear imperative now to maximise the use of that data captured at such great expense. The information society in which we live makes it ever easier to discover and exploit that data, in principle. However, data collected in the past for a specific application is not necessarily directly re-useable in a new and different context. The original collection process will have imposed many restrictions and brought many assumptions which need to be re-evaluated against the new usage scenarios. More difficult still is the fact that many of the underlying assumptions and restrictions were not explicitly stated, for example in data specifications, at the time but are simply inherent in the data itself. In practice, this means that data often has to be qualified and tested for the specific criteria important to a new pattern of use. This paper discusses the requirements for a rules language which is capable of expressing such consistency criteria for spatial and other linked data which is initially found in scattered locations in an internet or intranet environment. The process of representing the knowledge inherent in scattered data and the rules which constrain it in a portable and platform independent way is illustrated through a series of examples. Knowledge persisted in this way can be used within real execution environments to generate automatically important data consistency and quality metrics. The storage of such quality metrics within a database and provision of registry enquiry services opens up the possibility of searching for data on explicit quality and consistency grounds to meet welldefined application needs. The implications of this for spatial data re-use and advanced data mining applications are reviewed.