CERIF (Common European Research Information Format) emerged first as a simple standard not unlike a library catalogue card or the present DC (Dublin Core Metadata Standard) and was intended as a data exchange format. It was based on records describing projects, with persons and organisational units as attributes. However, it was soon realised that in practice this CERIF91 standard was inadequate: it was too rigid in format, did not handle repeating groups of information, was not multilingual / multi character set and did not represent in a sufficiently rich way the universe of interest. A new group of experts was convened and CERIF2000 was generated. Its essential features are: (a) it has the concept of objects or entities with attributes such as project, person, organisational unit; (b) it supports n:m relationships between them (and recursively on any of them) using 'linking relations' thus providing rich semantics including roles and time; (c) it is fully internationalised in language and character set; (d) it is extensible without prejudicing the core datamodel thus providing guaranteed interoperability at least at the core level but not precluding even richer intercommunication. It is designed for use both for data exchange (data file transfer) and for heterogeneous distributed query / result environments. With CERIF2004, minor improvements in consistency have been released. With CERIF2006 substantial improvements have been implemented with the model, concerning in particular the introduction of a so-called Semantic Layer, that makes the model flexible and scalable for application in very heterogeneous environments. (Excerpt from this source)
↧