Where Do Topic Maps Come From?

ABSTRACT

Topic maps have attracted considerable attention since they were standardised by ISO in January 2000. Many papers have been written, several books are in the works, training courses are readily available, and major applications are already seeing the light of day. Once they have understood the basic concepts, people start to see topic mapping as the answer to some of the very thorny information and knowledge management problems that confront them - and they begin to cast around for the tools they will need to exploit topic maps. There are already a number of free and commercial topic map tools to choose from, and vendors such as Empolis, Infoloom, Mondeca, and Ontopia have mature commercial offerings consisting, for the most part, of engines and browser applications. However, the current paucity of industrial strength editing tools for topic maps is seen by many as a stumbling block. "Topic maps are all very well", they say, "but we just can't afford to have expensive editorial staff creating topic maps manually."

The aim of this presentation is to show that these kind of fears are largely unfounded. It demonstrates that topic maps can and will come from many disparate sources and that by no means all of them will be created directly by hand. In fact, most topic maps will probably not even be maintained as topic maps, but rather as structured information in some other form, such as databases, metadata sets, and XML documents. The presentation will discuss the relationship between schemas and DTDs on the one hand, and topic map ontologies on the other, showing that much of the knowledge engineering performed for one purpose can be reused for others. It will also show that as a consequence of this, much of the data that resides in existing systems can be harnassed for the automatic generation of topic maps. Two novel approaches for doing this -- Virtual Topic Maps and the Metadata Processing Framework -- will be described and demonstrated.

Topic maps that have been generated automatically can be enriched by the manual addition of more manageable numbers of topics and associations. The presentation will describe some of the ways in which this can be done without having to confront editors with angle brackets, and also present requirements for a user friendly editorial interface to topic maps. Finally the presentation will show in practice how the power of topic map merging can be exploited to bring together topic maps from such disparate sources into a single, unified whole.