Re: [ld4lt] META-SHARE ontology metamodel comments

Hi Tiziano,

Thanks a lot for the analysis! Could you record your findings in the
"discussion" column at [1] when you have the opportunity? In this way they
will be easier to process and follow.

Regards,
Jorge

[1]
https://docs.google.com/spreadsheets/d/15SE4_qAqYFostmD52uKxpkCPZh1f5TrPeoXKNTlDYpQ/edit#gid=0


2014-07-31 12:05 GMT+02:00 Tiziano Flati <flati@di.uniroma1.it>:

> Hi all,
>
> I've been analyzing the META-SHARE ontology metamodel and, after
> a thorough discussion with Paola and Roberto, we came up with some comments
> and results (see below).
>
> I also plotted the ontology as a graph (using a bit of shell
> and java programming plus a pinch of Cytoscape). Thanks to this I was able
> to easily visualize the backbone of the metamodel and its features.
> Please have a look at the attached zip file which contains the plotted
> graphs I produced (in png, svg and dot format) [uptodate to yesterday].
>
> *=== Statistics ===*
> Here follow a few statistics on the number of classes, individuals and
> properties contained in the spreadsheet.
>
>    - 63 core classes in total;
>    - 509 individuals in total;
>    - 84 relations in total.
>
>
>  *=== Comments ===*
> We report our comments, sheet by sheet: CORE-classes, CORE-individuals,
> CORE-properties.
>
> *CORE-classes:*
>
>    1. Thanks to the graphs produced, simple typos could be easily
>    spotted (such as foaf.Agent instead of foaf:Agent or ms:characterEncoding
>    instead of ms:CharacterEncoding) ;)
>    2. What is the purpose of swrc:ResearchProject? It is also unlinked
>    from the rest of the ontology.
>    3. bibo:Document has superclass foaf:Thing, while all other bibo:*
>    have foaf:Document as superclass.
>    4. Either keep bibo:Document or foaf:Document.
>    5. The ontology includes foaf:Organization and foaf:Person, but not
>    foaf:Group (but all the three are under foaf:Agent).
>    6. 24 classes have owl:Thing as their superclasses (among which: ms:LinguisticInformation,
>    ms:CharacterEncoding, ms:MultilingualityType, ms:LanguageCode, etc.) Shouldn't
>    it be more structured?
>    7. ms:AnnotationType should have ms:LinguisticInformation as superclass
>    8. I cannot find ms:modalityAnnotation in the schemata (but I can find
>    its 7 individuals)
>    9. We seem not to find any quality-related annotation type (we though
>    see ms:SyntacticAnnotation, ms:SpeechAnnotation, ms:SemanticAnnotation,
>    ms:ModalityAnnotation, among others)  Why don't we include something like
>    ms:QualityAnnotation?
>    10. ms:LexicalConceptualResource is not structured at all: ms:WordNet
>    and ms:FrameNet are direct subclasses (while they should at least be under
>    ms:ComputationalLexicon). Besides this, the choice of WordNet and FrameNet
>    as resources to include are arbitrary, to say the least (we suggest
>    including BabelNet, DBpedia, YAGO, etc.). We suggest something more
>    structured, such as introducing ms:MultilingualComputationLexicon and
>    putting ms:ComputationalLexicon under ms:Lexicon
>
>
> *CORE-Individuals:*
>
>    1. 38 nodes are isolated. These include things like ms:annotationFormat,
>    ms:annotationMode, ms:tagset. This is probably not necessarily bad,
>    but should be avoided (at least choose owl:Thing as super type).
>    2. We propose to change ms:Princeton_Wordnet
>    into: ms:Princeton_Wordnet_license
>    3. We would call "ms:Availability" something more like
>    "ms:LicenseUse". In particular, we don't see any particular advantage in
>    keeping this class along with the ms:restrictionsOfUse; it seems there is
>    potential conflicting information here (for example, ms:notAvailableThroughMetaShare
>    and ms:onlyMSmembers seem somewhat related and/or redundant). This should
>    be moved under the license module, anyway.
>    4. "letter" seems to be a missing value for ms:SegmentationLevel
>    5. Cannot find ms:StandardsBestPractices in the MetaShare schemata
>    (while I can find ms:conformanceToStandardsBestPractices). Anyway,
>    this class should be called something like "ms:resourceFormat", since it
>    involves formats like GrAF, XLIFF, OWL, RDF, SKOS, TMX, etc.
>
> *CORE-properties:*
>
>    1. Over 84 relations, 35 have non-empty domain or range;
>    2. Only 28 relations are well-defined;
>    3. 7 relations have empty domain;
>    4. Cannot find ms:UserManual in the Meta-Share schemata.
>    Anyway ms:UserManual and ms:Documentation seem to overlap.
>
>
> I based my analysis on the basis of the following links:
>
>    - Support community:
>    http://www.meta-share.org/portal/knowledgebase/CreationMode
>    - Usage statistics: http://spraakbanken.gu.se/metashare/stats/usage
>    - Schemata:
>    https://github.com/metashare/META-SHARE/tree/master/misc/schema/v3.0
>    - Git project: https://github.com/metashare/META-SHARE/
>    <https://github.com/metashare/META-SHARE/tree/master/misc/schema/v3.0> (clone
>    at https://github.com/metashare/META-SHARE.git)
>
>
> Hope this contributes to further discussions : )
>
> Best,
> Tiziano, Roberto, Paola
>



-- 
Jorge Gracia, PhD
Ontology Engineering Group
Artificial Intelligence Department
Universidad Politécnica de Madrid
http://delicias.dia.fi.upm.es/~jgracia/