Associations and Conceptual Topic Maps

In the CTM ontology, world-views for Conceptual Graphs are cast into abstract graph structures for XTM 1.0   Their related PSI-sets define an open XML dialect of very expressive association types, to which people and software can both relate.

CTM is a rich, practical, wire-level format for modeling things, properties, thoughts, and processes at near-human levels of abstraction.  Based on English grammar forms, it is useful in natural language processing and formal semantic modeling.

Its associations are versatile structures of topics.  Each can easily represent one CG instance, such as this 7-topic example which models a simple Go event: 
[Person: Romeo]<-(Agnt)<-[Go]->(Dest)->[City: Mantua]
In XTM and CTM, the topics directly on both sides of the Go association type would generally be called role types.  The topics at each end are generally called their role players  (Here they have typing topics preceeding a ':').  Role player topics can play other roles in other associations simultaneously. In fact, the set of all roles it plays is (besides its own type) THE main characteristic of every topic .

It is often hard to nudge two complex formalisms into compatibility, but as the example above helps illustrate, Topic Maps and Conceptual Graphs have plenty in common to make this attempt seem inevitable

Their relative strengths and weaknesses also make it very worthwhile, because even though CTM association types will seem familiar to both user communities, a merged  formalism offers more modeling horsepower than either parent alone can provide.

 

What CTM Brings to Conceptual Graphs

XML Persistence - CGs have been limited by having no standard interchange format.  By mapping them into XTM (a standard of IS0) and PSIs (a standard of OASIS), the CTM ontology can remove such mechanistic limitations.

Open Data Types - PSIs let CTM users define new types of Strings which CGs may use as linguistic entities - such as numbers, names, URIs, formulas, facet IDs, scripts, etc.  Validation constraints for each legal instance must be openly defined on a PSI-compliant public web page whose URL acts as data type ID.

Associations - The primitive role/relations of CGs, however elegant, are properly building blocks for larger graph structures.   XTM associations let CTM assemble these CG role types into extended graphs of formally typed topics which - in the same fashion as an English clause - can express a complete thought

Available Tools - By tapping XTM standards, CTM lets many existing tools (often open source or free downloads) be used to query and visualize Conceptual Graph structures of associated XTM topics.  These topic structures may form, contain, relate or extend any CGs desired.

Graph Templates - XTM association types, if properly constrained, also serve as blueprints for constructing their own instances on demand.  So applications can use them to auto-assemble Conceptual Graphs or instantiate prototype Parts for them, in nearly any modern coding or modeling language you wish to support.

 

What CTM Offers to Topic Maps

Upper Ontology - XTM defines clear graph structures, but it omits standards for what its graphs actually represent.  CTM provides a useful answer - CG-based models involving any type or instance of object, substance, situation, property or role commonly cited within English paragraphs.

Whole-Part Models - CTM patterns model subjects from the perspective of their composition, a human-like viewpoint which can complement and greatly enrich the classification tree most Topic Maps now use as their sole topic hierarchy 

Situation Models - With constrained case roles, CTM lets association types be defined that model any spatial or temporal relation, any event, or any change of state.  To instantiate such a model, one need only specify its type - named by an English verb or preposition - plus its timing and other known role player topics

Logic & Constraints - XTM does not express the truthfulness of its graphs, but CTM adds such a model.  Using constraints, Boolean types, and the truthfulness of subgraphs, it supports constraint-based validators for every CTM instance.  It can also trigger forward chaining templates embedded in platform-portable IF-THEN rules, which complement backward-chaining queries in Tolog, etc..

Bulk CTM association types are now being defined at Lexikos Corporation for use as an output format.  The input format is paragraphs in any controlled English grammar or semantic domain that our clients would like us to transcribe into CTM.

We will truly welcome constructive criticism from all interested firms and public groups regarding our PSI-sets on the CTM core, and (eventually) on our equivalent XTM 1.0 files, which will become freely downloaded once they stabilize.

Meanwhile, please contact me with any questions or suggestions on CTM association types, and most especially to discuss your needs in such areas, and any ways that bulk CTM types or instance graphs might help advance your missions.

And when you eventually decide to add controlled-English I/O modules for your web apps or enterprise software, please consider using our unique English lexical data, which can make any vocabulary-based R&D process simpler and far less costly.

- Dan Corwin