When one Logic is Not Enough: Integrating First-order Annotations in OWL Ontologies

Tracking #: 3440-4654

Simon Flügel
Martin Glauer
Fabian Neuhaus
Janna Hastings

Responsible editor: 
Guest Editors Tools Systems 2022

Submission type: 
Tool/System Report
In ontology development, there is a gap between domain ontologies which mostly use the web ontology language, OWL, and foundational ontologies written in first-order logic, FOL. To bridge this gap, we present Gavel, a tool that supports the development of heterogeneous 'FOWL' ontologies that extend OWL with FOL annotations, and is able to reason over the combined set of axioms. Since FOL annotations are stored in OWL annotations, FOWL ontologies remain compatible with the existing OWL infrastructure. We show that for the OWL domain ontology OBI, the stronger integration with its FOL top-level ontology BFO via our approach enables us to detect several inconsistencies. Furthermore, existing OWL ontologies can benefit from FOL annotations. We illustrate this with FOWL ontologies containing mereotopological axioms that enable new meaningful inferences. Finally, we show that even for large domain ontologies such as ChEBI, automatic reasoning with FOL annotations can be used to detect previously unnoticed errors in the classification.
Full PDF Version: 


Solicited Reviews:
Click to Expand/Collapse
Review #1
By Torsten Hahmann submitted on 31/Jul/2023
Minor Revision
Review Comment:

My apologies for taking so long to review the second iteration of the paper. I think the paper is now in a better shape, though some explanations in the paper are difficult to understand. Addressing these would improve the overall readability of the paper.

The most critical of my comment concerning section 4.3 has been addressed by the authors. However, the added text and the intro to the evaluation could be a bit clearer. In particular, the authors stress that they evaluate whether FOWL can capture "ontologically meaningful axioms". I have no idea what is meant by that - it seems like an awkward (and unnecessary) emphasis. Presumably, all axioms that one wants to express are ontologically meaningful - otherwise one wouldn't care. I would suggest removing that phrase and rewording lines 1-8 and 24-26 on p. 7 and l. 40-41 on p. 9 ("ontologically meaningful axioms and reasoning tasks")

Other comments/suggestions:

9:43, you should you stress here already that these axioms are automatically derived from from SMILES annotations

3:42: "More specifically, a FOWL ontologiy is an OWL file, ..." The remainder of the sentence is confusing. What do you mean by "selected" here? Please clean this up.

4:35-28: This sentence is difficult to parse and also vague. Somewhere later on, you say you use Levenshtein distances to do name mapping. Please be more clear here.

8:34: "This is especially useful ..." - Maybe reword to "This is, for instance, useful ..."

9:47: delete "a" before "biologically relevant"

10:10: delete "an" before "initial experiments"

11:30-35: These two sentences are really difficult to follow, especially the second sentence: ""It should, however, ... " Please rewrite this to make it more clear.

11:45: "conjecture is B(a)" -> "conjecture B(a) is"

11:49: "is not an axiom" -> "is not an axiom or theorem" (you're using the transitive closure of subsumption)

12:23-29: the notion of unexpected proofs/counter-examples in ontology development has been explained in detail in Katsumi & Gruninger: https://pdfs.semanticscholar.org/30b6/5dc0a0b9196d07214c2ef0edd1b1515824...
You should probably refer to that work for details

Your response to my remarks #7 and #9 make sense, but I couldn't find the place where this is addressed in the paper. I expected it in Section 2.

Review #2
Anonymous submitted on 14/Aug/2023
Review Comment:

My previous comments were minor and I see most of my concerns addressed. The authors added a reference for the mapping they used and provided a link to the implementation (this latter was already present in the previous version). They did not address the issue that their approach does not really follow the semantic web idea, in the sense that they do not employ a triple syntax and do not use uris to name their concepts, but they at least explained the reason for their choice which is that they mainly wanted to have a notation which does already work in Protege and that is only possible if the logical extensions are provided via annotations. I would, however, have liked to see a comment that a proper representation of the FOL extension could be future work. But as I think that this remark is minor, I recommend to accept the paper even without this addition.