Review Comment:
Thanks to the authors, the paper has really improved, but I still need some explanations and modification of the modelling choice and final ontology. Without understandable natural language definitions the final ontology is not reusable.
-------------------------------------------------------------------------
Methodology
The authors improve the description of the reusing ontology design pattern scenario. The source mentions in the article https://www.w3.org/2006/07/SWD/SKOS/skos-and-owl/master.html presents 3 solutions of SKOS and OWL combination. The article should contain a description of the pattern reused in the GLOSIS ontology. This information should also be added in the wiki. When reading the article, I understand that the third pattern “Formal / Semi-Formal Hybrids -- Part OWL, Part SKOS” was reused, build a graph of OWL classes and a graph of SKOS concept scheme instances (and related SKOS concept instances) linked by rdfs:seeAlso. Moreover the new OWL class is defined a subclass of SKOS concept class. Note that the article extends a little bit the pattern by adding SKOS concept scheme instances and subClassOf links. Thus, it is part of the novelty of the proposition that should be explained and argumented. Why adding a SKOS concept scheme instances and subClassOf links helps in some extends? To my point of view it is a way to duplicate the information and to create more errors. It is not easy to handle duplicate information.
----------------------------------------------------------------------
Modeling problem
due to the problem in Widoco documentation I have a look to some URI that are defined as individual and as OWL class.
glosis_pr:soilClassificationUSDAProperty a sosa:ObservableProperty ;
rdfs:subClassOf glosis_pr:soilClassificationProperty ;
rdfs:label "soilClassificationUSDAProperty"@en ;
skos:definition "ISRIC Report 2019/01: Tier 1 and Tier 2 data in the context of the federated Global Soil Information System. Appendix 1"@en .
glosis_pr:soilClassificationProperty a sosa:ObservableProperty ;
rdfs:label "soilClassificationProperty"@en ;
skos:definition "ISRIC Report 2019/01: Tier 1 and Tier 2 data in the context of the federated Global Soil Information System. Appendix 1"@en .
It is not possible to use a subClassOf link between two individuals, unless the GLOSIS ontology is an OWL Full one. Thus the individual should also be typed by owl:Class.
This type of modeling is named “pruning”. I would need explanation in the article about this choice. The paper does not mention this modelling choice.
-------------------------------------------------------------------------
State of the art
SOTER ISRIC model is reused by the GLOSIS ontology. The state of the art should contain a description of this model.
-------------------------------------------------------------------------
Improve definition of GLOSIS ontology element
The documentation of the ontology has improved but it is not sufficient. No understandable definitions are provided. The definitions contain only the source of information.
For example “ISRIC Report 2019/01: Tier 1 and Tier 2 data in the context of the federated Global Soil Information System. Appendix 1.” is the definition of glosis_cm:SoilDepthBedrock class.
glosis_pr:GL_Profile and glosis_sp:GL_Plot share the same definition "Data model for the Global Soil Information System (GloSIS) v1.0, edited by Tomas Reznik and Katharina Schleidt". This definition does not help to understand the classes.
Note that source of information seems to be reusable without restriction. So each class should be defined by an Aristotelian definition. SoilDepthBedrock is a type of sosa:Observation….
See Robert Arp, Barry Smith and Andrew Spear: Building Ontologies With Basic Formal Ontology, MIT Press, 2015 for example of definition.
A definition should explain the position of the ontology element in the hierarchy (precise its ascendant) and explain the differences with its siblings.
Note that in section 3.4.1 of the paper the authors provide the definition of main classes: siteplot, profile etc…
To resume due to the lack of natural language definition in the GLOSIS ontology the final result is not easy to use.
-------------------------------------------------------------------------
Figure 2
The figure 2 is too small to be readable. The image should be part of the wiki.
-------------------------------------------------------------------------
Figure 3
What is the graphical language used in the schema? I recommand to use CHOWLK language. The schema needs some legend.
The link between the NitrogenTotal and NitrogenTotalValue classes should be hasResult.
I assume that the two links between , means that the any instance of NitrogenTotal could be linked by the object property sosa:hasFeatureOfInterest to instance of GL_Layer or GL_Horizon. The two links are hard to understand without legend.
The figure should contains more information:
The sosa Procedure class
The glosis_lh:PhysioChemical , she contains some of the restrictions.
The glosis_lh:PhysioChemicalValue class
The qudt related class
The related SKOS concept scheme instance
I would appreciate that the figure 3 express the complexity of the proposed modeling.
-------------------------------------------------------------------------
Figure 4
The meaning of the link between classes are not explained. As far as I understand the ontology network some links express subClassOf other equivalentClass. I would appreciate that the two type of links are represented in a different way.
-------------------------------------------------------------------------
Table 1
Table 1 should be updated with prefixes presented at lines 41 to 51 of page 8. The lines should be removed and replace by reference to table 1.
-------------------------------------------------------------------------
Application section
https://w3id.org/glosis/open/LUCAS/topsoildata/ link answers “The requested URL was not found on this server.”
https://w3id.org/glosis/open/srdb/ link answers “The requested URL was not found on this server.”
the git https://github.com/bpbond/srdb is accessible.
-------------------------------------------------------------------------
Check widoco generated documentation
In the widoco documentation, the name pattern of ontology element are not coherent.
Example of different classe names in surface module :
• color wet
• SaltCover
Example of named inidvidual in surface module :
• SaltCoverProperty
• saltPresenceProperty
For example, in the documentation of SoilDepth class, Sosa:Observation is written “observation”. In GLOSIS and SOSA ontologies, the class name start with a Capital letter. A prefix should be added to identify the ontology source.
-------------------------------------------------------------------------
Check the iso 282580 ontology ttl file
The link to the iso 282580 ontology does not work
https://w3id.org/glosis/model/iso28258/2013/Surface
answer “The requested URL was not found on this server. Apache/2.4.29 (Ubuntu) Server at w3id.org Port 443”
the link to the iso standard does not work
http://standards.iso.org/iso/19156/ed-1/en/
answer “The page you're looking for has either moved or no longer exist.”
https://def.isotc211.org/iso19156/2011/SamplingFeature#SF_SamplingFeature
answer “The requested page could not be found.”
Most definitions of iso 28280 ontology class are well written. The iso ttl file should be checked there are some errors in element definition like:
iso19156_SF:SF_SamplingFeature
a owl:Class ;
rdfs:label "SF_SamplingFeature" ;
skos:definition ".lt.ul.gt. .lt.li.gt..lt.b.gt.Role of sampling features.lt..sl.b.gt..lt..sl.li.gt..lt..sl.ul.gt.Sampling features are artefacts of an observational strategy, and have no significant function outside of their role in the observation process. The physical characteristics of the features themselves are of little interest, except perhaps to the manager of a sampling campaign.EXAMPLE A .ldq.station.rdq. is essentially an identifiable locality where a sensor system or procedure may be deployed and an observation made. In the context of the observation model, it connotes the .ldq.world in the vicinity of the station.rdq., so the observed properties relate to the physical medium at the station, and not to any physical artefact such as a mooring, buoy, benchmark, monument, well, etc. NOTE A transient sampling feature, such as a ships-track or flight-line, may be identified and described, but is unlikely to be revisited exactly. A sampling feature is intended to sample some feature of interest in an application domain. However, in some cases the identity, and even the exact type, of the sampled feature may not be known when observations are made using the sampling features. .lt.ul.gt. .lt.li.gt..lt.b.gt.Classification of sampling features.lt..sl.b.gt..lt..sl.li.gt..lt..sl.ul.gt.A small number of sampling patterns are common across disciplines in observational science. These provide a basis for processing and portrayal tools which are similar across domains, and depend primarily on the geometry of the sample design. Common names for sampling features include specimen, station, profile, transect, path, swath, and scene. These are classified primarily by their topological dimension. Point-located samples are defined in this Clause, extensive samples of various dimensions in Clause 9 and specimens in Clause 10." ;
what means “ lt.ul.gt. .lt.li.gt..lt.b.gt” string
iso19156_OB:NamedValue
a owl:Class ;
rdfs:label "NamedValue" ;
skos:definition "The class .lt.i.gt.NamedValue.lt..sl.i.gt. provides for a generic soft-typed parameter value. NamedValue shall support two attributes. http://standards.iso.org/iso/19156/ed-1/en/" ;
the ttl file contains tags
iso28258:ProfileElement
a owl:Class ;
rdfs:label "ProfileElement" ;
skos:definition "Profle element is an abstract feature type grouping layers and horizons. Profle element may be considered as a horizontal feature that is parallel to the earth surface and that is part of the profle." ;
replace Profle by Profile
The object properties are not associated to a natural language definitions.
I would recommend the authors to check their file using OOPS or any checking tool.
-------------------------------------------------------------------------
Spelling errors
page 4 line 33 doamin → domain
page 4 line 50 remove “to the this domain”
page 10 line 49 “Then, each code-list value is modeled as an individual of type: the defined class and skos:Concept,” I found strange “of type:”
page 11 line 6 and 7 quadt → qudt
page 17 line 24 footnotehttps://github.com/bpbond/srdb
after page 18, the page numbering restart page 0
|
