Review Comment:
This ontology paper presents an ontology of government decisions and actions, in the context of Greek public administration. This is the revised version of an earlier submission.
Overall, the paper has been improved, but still several issues remain, and they are not all minor. Which leads to my suggestion of again doing a major revision of the paper.
First of all, I cannot assess the resource itself, since the link provided on the front page of the paper is broken. When following the redirect, and checking under the groups “ontology” pages, I cannot find any link to this project or the ontology unfortunately. This is also an example of a bad practice when publishing an ontology, since the ontology should have been given a permanent URI (e.g. through w3id) before publishing it, so that that URI can simply be redirected if the pages/files need to be moved. Also other links later, e.g. some of the links in section 4.5 about URIs are broken.
Section 2.1 has been improved and is on the right track, but needs to comment more on each ontology to make it clear exactly what is missing in each one for covering the whole set of use cases, and what this work adds to that existing work.
The relation to the previous Diavgeia ontology is still unclear in the paper. The new paragraph added on page 6 is unclear, details should be provided on the comparison between the new and “old” version. In addition, this paragraph seems to contradict the first paragraph of the methodology section later on the same page, where it sounds as if the “new” version is merely an extension of the previous one - but then it would still be monolithic etc?
Regarding the CQs of use case 3, all CQs seem to be aggregates - does this imply that the actual data does not need to be stored, but only the aggregates? Or is the list simply incomplete?
Section 4.1 is strangely placed. Is this the motivating scenario from which CQs have been generated? If yes, move it earlier, before the use cases and CQs. However, if this is a use case scenario for evaluation/application it should instead be moved later in the paper, after the ontology has been presented. Given that it mentions concrete properties in the ontology (which is not yet introduced), it seems it should be the latter. But given 4.2 the method description in 4.2 it seems it should be the former. In summary, this section needs to be moved and probably rewritten. Section 4.2 about the methodology is now also very brief, and does not seem to add up with the use cases and CQs in the previous section. If the ontology is constructed based on analysing an annotating documents, then how are the use case and CQs actually used in the process?
Fig 2 that illustrates some concepts of the ontology does not seem to be standard UML, and the legend is still very much unclear. It is still not clear what some of the relations mean, e.g. what does it mean that an object property connects two classes - is it domain and range or another axiom? Colors are mixed, and sometimes properties are shown in the boxes, but sometimes instead subclasses are shown, without any obvious explanation.
On page 11 the authors state that the ontology is composed of vocabularies of “mixed formats”, so how is your ontology represented then? Later is sounds a if it is an OWL ontology, but how do you then do this “mixing” in practice? The authors need to describe more in detail how the integration is done, e.g. through imports, or referencing URIs? But some vocabularies are not in OWL, e.g. the authors say that for instance CPOV is in RDF, so how is that possible? Additionally, what does “integration of controlled vocabularies” mean technically?
Overall the description of the ontology is not sufficient for an ontology paper. Please see [1] for the suggestion of a minimal set of things to describe! Since the ontology is also not accessible online, the quality of the actual artefact cannot be properly assessed.
In section 4.5 I am not sure what the authors mean by saying the “standard XML representation of a URI”, how can a URI be represented in XML? It seems to be more some data snippet about the thing identified by the URI? But why XML and not RDF if you are using linked data principles and RDF as stated both earlier and later in the paper?
The query of use case 3 on page 18 seems wrong. It does not make sense to count the documents and then display also the names and posts, since there will be several names and posts if the count is more than 1. It only works since there seems to be only one post in the data.
Further, it is unclear if the inferences described in section 6.2 are really used in the application of the graph? Or are these only examples for verifying the ontology structure? If this is only for verification, then what reasoning is actually used in the application of the ontology?
Last paragraph of section 6: Are those example instances? Then it does not seem fair to include them at all in the evaluation, since example data should not be considered part of the ontology. If not, i.e. they are actually part of the ontology, then what does it mean that they are not fully covering the classes?
Finally, the paper needs some additional proof reading. For instance, on page 8 the acronym CPV is used in a strange way in the sentences, as if it meant something else, e.g. a type in that vocabulary?
[1] Matentzoglu, N., Malone, J., Mungall, C. et al. MIRO: guidelines for minimum information for the reporting of an ontology. J Biomed Semant 9, 6 (2018). https://doi.org/10.1186/s13326-017-0172-7
|