A small celebration. I succeeded in integrating three tools to some extent. The pundits are:
- tOKo: a tool that supports the definition of ontologies from text documents. tOKo is an acronym that stands for "from TOKens to Ontologies". Work on tOKo is a significant part of my research.
- Sigmund: the "electronic shrink" I blogged about previously. Most of the Sigmund implementation heavily relies on tOKo.
- Triple20: probably the best RDF / Semantic Web editor in existance, written by my colleague Jan Wielemaker. More info on Triple20.
The result of integrating these three tools is really nice. Using tOKo one can drag-and-drop "terms" from documents and identify them as concepts (or classes) which can subsequently be edited in Triple20. Sigmund has a certain fancy for picking up patterns in texts and these patterns can then be refined further in either tOKo or Triple20.
Applications. We are using the above set of tools in a project I cannot blog about. A bloggable alternative is to apply it to the Abstracts of ECITE 2004 as suggested by Jack Vinson in Topic Maps: when you cannot be there. Gabriela Avram kindly mailed me these abstracts in a tOKo/Sigmund processable format. Unfortunately, my knowledge of the ECITE subject matter is rather limited (and volunteers, except for Sigmund, were hard to find). So, I post two screendumps of tOKo with some help of Sigmund. The one on the left is based on the full ECITE abstracts, the one on the right is based on the same abstracts omitting the "Keywords" inserted by the authors.
 |
 |
Spot the difference!
Potential. Reflecting a little bit. The Semantic Web community is struggling to find good applications for their inference machinery, whereas others are questioning whether ontologies are a good idea at all (e.g. Lilia Efimova in: Converging metadata and emerging ontologies). I think Lilia, and others, are right in stating that inferencing does not go very deep in "informal domains" (perhaps as exemplified by ECITE). In contrast, many technical domains benefit from ontologies. Integrating tOKo (which is ontology oriented) and Sigmund (which is socially oriented) seems like a starting point for further investigation.
Comments