Skip to content
« A Microsoftish Perspective on Standards
ECIM 2008 Presentation »

SSSW08: NeOn Methodology for Ontology Specification

You probably haven’t noticed (until perhaps now?), but it’s been quiet around here for a while. The blog has been at the bottom of my prioritized “tasks”. Nevertheless, I have been meaning to post about the NeOn methodology (and other stuff from SSSW’08 (coming)) for an even longer while.

In many ways, these are notes from my stay at the SSSW’08.

You can download the ontology specification document at the SSSW’08 web site.

NeOn Project

The NeOn methodology for Ontology Specification (NMOS) was developed as part of the NeOn project.The NeOn project is a $14.7M project funded by the EU (like all other research projects dealing with the Semantic Web, I guess, thanks to the Bush administration).

Among other things (such as lots of PhD projects and other research), NeOn has developed a toolkit; the NeOn toolkit, which includes a not too impressive ontology editor.

We were also introduces to the Watson semantic web search engine plug-in for said toolkit — also not too impressive, especially when thinking of usability. But then again: is there any good way for humans to interact with ontologies? See Enrico Motta’s presentation from the SSSW’08.

NeOn Methodology for Ontology Specification

The NeOn Methodology for Ontology Specification is said to ensure that the ontology that is being developed as part of a project is of high quality and satisfies its intended uses.

The NMOS specifies a process to follow when, or prior to, developing ontologies. According to NMOS, the ontology development process should start by identify and define:

  1. Purpose, Scope and Level of Formality
  2. Users
  3. Uses
  4. Requirements (Competency Questions)
  5. Group Requirements (categorization of CQs)
  6. Validate Requirements
  7. Prioritize Requirements
  8. Extract Terms

In the first step the team decides the purpose and scope of the ontology. That means identifiying the main goal and the granularity of the ontology to be produced. Also, NMOS suggests deciding the level of formality here (from natural language to rigorous formal). Topic Maps would probably be considered semi formal, while FOL is rigorous formal.

During step two, the end users are identified. In most real world cases, these would probably have been identified prior to even starting the ontology development process. It is, however, important not to loose sight of who should be using the ontology, so that the outcome of the process is something useful. Just like you’d appreciate personas and use case scenarios in any other kind of software project…

Usage cases / scenarios are developed as part of step 3. “How will the ontology be used?” is a given question here.

The requirements are gathered in step 4 through creating a list of competency questions (CQs). This list should cover every possible case, and will later serve as a checklist for whether the requirements are valid or not. Competency questions may also be used to test the end result (ontology).

The next step, step 5, of the ontology specification process is to group, or categorize, the CQs into groups of related questions. Example (job seeking ontology): all questions related to registering a CV in one group, questions related to job searching in another group, etc.

The validation in step 6 includes checking for missing or contradicting CQs, check for ambiguity, conciseness, etc.

Step 7 sees to that the ontology development team prioritizes the requirements across and within groups of CQs. The reason for prioritizing the requirements, is to help the team during the development process.

The concepts (classes), relations and attributes to include in the ontology are then extracted from the CQs in step 8. This is done by extracting the most used terms from the list of CQs and their answers. The answers to the CQs are also used to create terms for populating the ontology.

After completion of step 8, the ontology development team implements the ontology using the results from the ontology specification process.

Making use of the NMOS

One of the tutorials / hands on sessions at SSSW’08 was to form groups and develop an ontology based on / using the NMOS. It was quite interesting, but I got the feeling that the fact that the ontology creation process is iterative was given too little emphasis. Surely, your work will not be complete after checking off the last item in the above list.

Also, even though many of the steps would be part of any ontology development process, I felt it was a bit tedious to follow a specification like the NMOS. But that might be me, and not necessarily a good thing.

In any way, it is good to have a thought-through “pocket guide” to ontology specification.

As mentioned at the top of this entry, you can download the the ontology specification document at the SSSW’08 web site and have a go (or at least read) of it yourselves.

This entry was posted on Tuesday, October 7th, 2008 at 19:35 and is filed under Semantic Web. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Add Comment