topicmap.com

Hand-crafted Machine-generated Knowledge Interchange

Getting Started

This is a bare bones guide to getting a prototype up and running from scratch.

1. Understand the topic map data model and basic concepts

The introductory and general articles will give you the basic concepts of the topic map model. The XTM 1.0 specification also contains some helpful prose explaining the model before explaining the syntax.

2. Become familiar with the XTM syntax

The XTM 1.0 specification is the obvious place to start for understanding the interchange syntax :) The specification gives many examples to illustrate the concepts discussed. The topic map tutorial at Bond University, covers the basic concepts and XTM syntax and provides a good overview of the main constructs.

3. Information design

Topic map literature has been mainly focussed on the layer below the ontology and so there is relatively little topic map literature devoted to information design. The topic map data model has been designed to be flexible and generic enough to express other ontologies.

You may have already modelled your data or there may already exist an ontology in your particular domain. These will provide useful starting points. In brief, it is necessary to define the various entities and relationships you wish to model. Recommended reading:

Knowledge Representation: Logical, Philosophical, and Computational Foundations (Sowa)

Building Enterprise Information Architecture: Reengineering Information Systems (Cook)

4. Hand craft "template" topic map

Topic maps can be merged to provide a unified view of a number of maps. During the merge process topics may be merged if they have the same name in the same scope, or if they represent the same subject. It is therefore possible to design maps which work closely togther when merged.

A popular pattern is to design a "template" topic map which captures the basic structures. This map will hold the various "typing" topics and is usually under the control of the ontology designer. The entity and relationship types defined in the previous step make up the template. These include: topic types, association types, association role types, occurrence types and themes.

A second, "production" map can then be used by topic map editors to add content to the template map. The production topic map makes use of the constructs defined in the template topic map.

5. Test template in navigator

Once the template topic map has been designed it can then be tested in a topic map navigator. This will allow debugging for both well formedness and syntactic validity. Once the map has been successfully loaded it can then be more easily checked for semantic validity.

The respective vendors each have topic map navigators. Ontopia has an free Navigator limited to 5000 topics, associations and occurrences and unlimited by time. The Navigator requires the Java 2 SDK be installed. Viewing a topic map is a simple as dropping the map into a directory for browsing.

6. Machine generate "production" topic map

A number of approaches can be taken to generate the production topic map. One approach is to publish an XTM instance from some other persistent data store such as a relational database containing metadata. A more interesting approach is to "extract" metadata and relationships from a set of structured or unstructured documents. Making implicit relationships explicit within a topic map allows users to easily access knowldege which may have remained obfuscated. The Metadata Framework will aid the metadata extraction process.

Note that the template and production maps should be kept to under 5000 topics. This is done for two reasons. (i) Evaluation software is typically limited to this amount (ii) An XTM instance must be held in memory.

7. Test in navigator

A mergeMap element in the template topic map will merge the production map with the template when the template map is parsed. If all goes well you should be able to navigate both of the maps as if they were one. Well done.