GrahamMooreChief Technical Officer

STEP UKConsultingUnit BDorcan ComplexSN3 5HQSwindonUK+44.1793.485465+44.1794.485451gdm@stepuk.comhttp://www.stepuk.com

Graham Moore is Chief Technical Officer for STEP UK and is responsible for the development of skills and technology that are used in the construction of information systems. Graham is the development lead responsible for the realisation of the X2X XLink product from STEP UK and has been developing Topic Map software for the past twelve months. Graham has worked in the SGML and XML industry for several years, including working for Database Publishing Systems Ltd. Graham graduated from Southampton University with first class honours degree in Compiter Science. He is now studying for a Phd in Distributed Information Management at Southampton University, where the current focus is on the symbiosis of metadata and linking. topic mapsISO/IEC 13250:2000semanticinterchange A problem has arisen with the proliferation of XML languages within and across domains the problem of how to successfully interchange information in a way that is semantically correct. This paper looks at existing tools for semantic interchange such as Architectural Forms and then continues by presenting an innovative use for the new Topic Map Standard. In this paper we illustrate how the Topic Map standard can be used for the semantic interchange of structured information. Unlike Architectural Forms it does not require the 'structured form' to be SGML or XML due to the powerful addressing mechanism that Topic Maps inherits from HyTime where by any structural form can be used. This paper concludes with a synopsis of how the application of this technology could finally enable reliable and trusted interchange of information in the emerging eBusiness and eProcurement industry.

This paper proposes an innovative use of the Topic Map standard and supporting implementations that facilitate the reliable semantic interchange of structured information. The basic premise is based on the premise of using two levels of agreed metadata. The two levels are the Topic Map abstraction and the schema abstraction. The schema level has been around for a long time and is a description of how an instance must be structured in order to be valid. An example of schema languages are DTDs and Java Classes. However, to achieve semantic interchange using the schema level metadata it is necessary to map every construct from one schema into the corresponding structure in another. This results in a star effect where, as the number of schemas increases, the number of mappings increases by the number of existing schemas. The schema to schema approach is not a scalable and effective model for the semantic interchange of information. Therefore the introduction of another layer, i.e. the Topic Map layer enables a more scalable, generic and open solution. By using the topic map to capture the essence of what a particular schema structure is, e.g. a first name, and then defining Topic Occurrences that map to schema constructs we can provide the association between different schemas. Unlike the schema-to-schema model it is not necessary to map between every structure, only from a new schema construct to the Topic that represents its 'isness'. An interchange application can, given an instance, move to the schema construct for a given instance part and then from the schema traverse to the Topic Link of which it is an occurrence. From the Topic it is then possible to select other occurrences that reference schema constructs in alternative schemas. A generic schema type interface(i.e. one for DTDs, one for XML Schemas) can then transform the structure into a form understood in the new schema. Further sections of this paper will show how associations can be used to express the dependency between schema constructs, how public subjects and repositories will enable disparate communities to define their schema with regard to existing schemas. Finally it will demonstrate a technology that implements the features described in this paper. Ultimately, this paper presents the application of the Topic Map technology to solve aguably the most fundamental problem facing information systems in todays open-information world.