Review Comment:
This article presents an extension to an existing ontology for describing medieval manuscripts, along with a demonstration of its application through an RDF export from different sources and a web interface/portal. The focus is on integrating existing digital collections of legal manuscripts, specifically the projects IRNERIO and Mosaico, by building on the existing MMDIO ontology which itself builds on other ontologies such as FaBiO and FRBR, as well as standard vocabularies including DCTerms and FOAF. The concept of a Polymorphic Knowledge Graph also plays an important role in this discussion.
This is a revised version of the article, and the new version is significantly improved. It is clearer regarding the goals and new contribution compared to previous work, and regarding the limitations that are inevitable in any such project. Overall it is sound and close to being ready for publication. The comments that follow are long, but they all refer to minor points that I think can be addressed quickly and easily.
The article is generally well-written, clear and with few obvious errors or typos in the text. The data and documentation are available in a standard manner (GitHub in standard formats). However, the bibliography contains a large number of errors which suggest that it was automatically generated but not checked (reference [2] is from "Journal Name is Missing", capitalisation is often wrong [3, 6, 19, 23], and so on.
The new conception of the "Polymorphic Knowledge Graph" seems very intersting and does appear to be the principal new contribution of the article, as the authors note. However, barely two pages are dedicated to describing it (Sections 5.1-5.4, since the rest of Section 5 seems standard RDF engineering), and it remains very unclear to this reader how the PKG really works or what its real contribution is compared to a "standard" Knowledge Graph. An example is given that Authenticum can have different representations (a manuscript, a digital version, a physical book), but it is not stated whether this is at the ontology level (the same object is modelled as class Book, Manuscript, ManuscriptCollection, for instance, as polymorphism would normally imply), whether the one record in the original RDB is transformed into different triples with different classes, or something else. The examples seem to suggest the former, namely that Authenticum is "a single manuscript instance' (so an instance of class Manuscript).
There remains a lack of clarity to this reader around key terms such as 'manuscript', 'book', 'manuscript collection', 'work' and so on. For instance, it is stated that Authenticum is 'a single manuscript instance' which links to 'codices' that represent 'manuscript collections', and that this preserves the relationsbip between the 'work', its manuscript representations and their respective collection. This implies that 'Authenticum' is an instance of class Manuscript (which is itself a subclass of Expression), but also that it is a 'Work', unless 'work' here is not meant strictly in the FRBR sense of the term. Similarly, 'Manuscript Collection' does not seem to be defined anywhere, while fabio:Book seems to be used in practice only for digital resources, if I understand correctly, even though memo:Manuscript is a subclass. Fig. 3 certainly helps, although it does not seem to include memo:Codex, and perhaps more or clearer explanation of concepts would help, particularly as some of the terms are used in ways that are not necessarily intuitive. The online documentation also helps but requires some effort to interpret.
Finally, there are a relatively large number of small points or suggestions, but these are very minor and could be fixed very easily:
The fact that the project is focussed on manuscripts in the context of legal studies should be stated more clearly at the start and/or in the abstract, rather than waiting until §3, otherwise the earlier passing references to legal studies seem out of place and make one wonder why legal studies are mentioned and not any number of other disciplines.
p. 3, end of introduction: It seems odd to this reader to exclude Section 3 from the summary, so I would suggest at least a few words, even though it has only just been mentioned in the previous paragraphs.
p. 5, Hebrew Manuscripts ontology "supports integration via tools like owl:sameAs": certainly, but is it not the case for all ontologies? I don't see that this statement adds anything.
p. 8, "using SAMOD methodology": a reference and/or definition is needed.
p. 9 "filling this gap by the development or utilisation of a formal ontology": OK, but this seems to me to slide between expressiveness and formalisation: the latter does not necessarily have any impact on the former.
p. 9 "The location associated with memo:Manuscript represents geographical metadata recorded for the manuscript and does not imply that the Expression itself is a physical object." While this clarification is welcome, it still isn't clear to this reader which metadata is being considered. Places in the text? The place of composition? The place of origin of the author? ... Similarly, statements such as "Figure 7 indicates that the manuscript was originally kept in Paris and is now preserved in Italy" (p. 25) seem to contradict this.
Section 4.2 seems out of place, since the preceding and following sections are much higher-level, whereas this is really in the details. Honestly I'm not sure if it's really necessary, or if kept more discussion seems necessary. Perhaps move it to an appendix?
Section 4.6.2 still seems very arbitrary to me. I am not at all familiar with the FOCA evaluation, but the decision that Q5 is at 75%, for instance, makes me wonder on what basis (and why not 72%, or 78%, etc.). As presented, the numbers seem very much "plucked out of the air", so to speak, and need more justification and explanation.
p. 19: "For example, XV century was translated into a start date of 1500-01-01T00:00:00 and an end date of 1599-12-31T23:59:59" OK, though this pragmatic move hides a lot of complexity that probably should be acknowledged (we all know that "XV century" is never intended to be precise to the second or even to the year, let alone does XV century begin in 1500 or 1501, which calendar, was it really 1 January, etc.). Similarly "enabling accurate time-based reasoning" seems to ignore that this is a false precision, and the accuracy is therefore questionable. This is probably unavoidable but should be acknowledged.
p. 21: "These use cases validate the PKG’s ability to perform cross-platform querying": But the examples seem to be rather harvesting data from different platforms and then querying the resulting dataset, rather than truly dynamic cross-platform querying, no? Similarly, on p. 22, "Manuscripts retrieved from both “Mosaico” and “Irnerio” digital collections", but in fact all the entries are "Collection: Mosaico".
p. 24: "In addition, the book’s author is ‘Cristina Vano’, which enhances the context and credibility of the work". This is not at all clear to me. In what way does the author enhance credibility?
p. 26: "Figure 7 provides information about the previous location, ‘Paris’, and the current one in ‘Italy’", but the identifier is "Paris, Mazar. 452" which is clearly the Mazarine library in Paris, not Italy. Is this a typo?
In general, the figures showing the knowledge graph are useful, but it would also be helpful to give a sample of results for each case.
SPARQL listings consistently have the typo "mauscript" for the variable "manuscript". Of course this is still valid code but would be more readable and professional if fixed.o
Section 6.2: Perhaps an aside not necessarily to be addressed, but since the whole point of this project is integration, I find it odd that the interface then explicitly separates Mosaico and Inerio back into separate sections, and true integration is only accessible via SPARQL.
|
