Reviews:PierluigiMiraglia about Term-based thesaurus to lightweight ontology – record-based model

From Odp

Revision as of 22:53, 7 September 2009 by PierluigiMiraglia (Talk | contribs)
(diff) ←Older revision | Current revision (diff) | Newer revision→ (diff)
Jump to: navigation, search


Review Summary: The pattern is described clearly and exhaustively, and includes relevant examples. It addresses a common problem in transitioning enterprises and applications from legacy resources such as thesauri, terminologies, etc. to more semantically robust data structures. As such, it is worth discussing in WOP2009 events; it is hoped that additional revisions and discussion contributions will lead to a consensus approach to this general problem.

I recommend revising and/or refining the following topics (see below for fuller description): - Use of OWL Full - Consider SKOS - BT/NT interpretation - UF interpretation - Illustrative flow charts

It may be enough, in most cases, for the author(s) or discussion participants just to clarify their position.
Reviewer Confidence: High with respect to the problem area the pattern addresses. Medium with respect to design patterns in general.
Problems: 1: Although labelled as 'lightweight' by the author, the resulting ontology is OWL Full. Is this as intended? This is a common issue in this context and seems worth clarifying. Most people don't have in mind OWL Full when considering lightweight ontologies.

2: Would it not be more natural to transition a term-based thesaurus to a SKOS formalization, and then map skos:Concepts terms to classes (or individuals) in an ontology? This may allow flexibility in dealing with the aforementioned OWL Full problem.

3: Is it typically desirable to turn BT/NT relations into transitive sub/superclass relations? This pattern, as written, assumes it is, regardless of the context or purpose of the intended deployment. Again, the SKOS specs proceed more carefully. It may be worth to consider not only an example of applying this pattern, but also a use case -- i.e. an example of why one might want to apply it.

4: In many thesauri the UF relation is used for synonymous term strings (i.e. synonyms in a synset) rather than self standing terms. It seems that would be a simpler solution, too.

5: The graphical charts could be clearer in representing iteration over the NT terms of a term created (rather than just say 'repeat', use a cyclic graph for that part of the chart).
Community Relevance: In my experience, the problem addressed looms large in the community, and is likely to remain central as more organizations adopt semantic technologies.
Relation to Best Practices: The solution presented is fairly natural, and I would guess pretty common. However, there are alternatives, and there should be more considerations of costs v benefits in regard to the alternatives as well.
Reusability: Highly reusable, in a variety of contexts and initiatives.
Relations to Other Patterns:
Overall Understandability: Excellent
Clear Problem Description: The problem should be described in more practical terms, without prejudging the options. I think it would be better to formulate the problem faced by the "consumer" of the resulting resources, rather than state that the result must be a lightweight ontology.
Clear Relevance and Consequences:
Clear Figures and Illustrations:
Missing Information:

Posted: 2009/9/8 Last modified: 2009/9/7

All reviews | Add a comment at the bottom of this page
Personal tools
Quality Committee
Content OP publishers