- From: Joćo Paulo Almeida <jpalmeida@ieee.org>
- Date: Thu, 12 Dec 2013 11:26:52 -0200
- To: Government Linked Data Working Group <public-gld-wg@w3.org>
Dear All, Here goes some additional comments on the best practices document: With respect to sections 6 and 7, I found their role a bit confusing in the current text. This is because some criteria for vocabularies in section 6 are repeated in section 7 (with a different purpose). I would like to suggest to separate ³Requirements for Vocabulary Creation and Dissemination² (which would emphasize what vocabularies should look like if you are creating them), from ³Additional Vocabulary Selection Criteria² (which would emphasize what you should consider if you are looking for an existing vocabulary to reuse). I would suggest to present first ³Requirements for Vocabulary Creation and Dissemination² (an inversion in the current order). In ³Requirements for Vocabulary Creation and Dissemination² we would say things such as: A vocabulary must be documented A vocabulary should be self-descriptive A vocabulary should be described in more than one language A vocabulary should be accessible for a long period A vocabulary should have persistent URLs A vocabulary should provide a versioning policy A vocabulary should be published following available best practices but not: Vocabularies should be used by other data sets Neither: Vocabularies should be published by a trusted group or organization These last two comments would be in Vocabulary Selection Criteria. Instead of repeating the criteria from ³Requirements for Vocabulary Creation and dissemination², the section on ³Additional Vocabulary Selection Criteria² would just refer to the : Make sure that the vocabulary was created following the recommendations aboveŠ and then add the other specific criteria concerning reuse (level of adoption, trustworthiness of publisher). (By the way "should² has not marked as special keyword in these lists.) For Section 7, I got confused with the 2nd box: "URIs for properties with non-literal ranges What it means: Name all properties as verb senses, so that triples may be actually read; e.g., hasProperty .² Is this in the right place? Is seems so unbalanced with respect to the other boxes. Regards, Joćo Paulo On 5/12/13, 12:02 PM, "Joćo Paulo Almeida" <jpalmeida@ieee.org> wrote: >Dear All, > >I have not finalized my review, but would like to provide my comments so >far (only up to section 5): > >There are some statements about the normative force of the "specification" >that do not seem to make sense if this is progressing towards a note. >(See, the first sentence in section 4.) > >However, I think that it still makes sense to use the keywords from >RFC2119 to define what conformance for vocabularies would mean. > >Currently, the text does not use this keywords and reads as follows: > >"A data interchange, however that interchange occurs, is conformant with a >vocabulary if: >it is within the scope and objectives of the vocabulary; >all classes and properties defined in the vocabulary are used in a way >consistent with the semantics declared in its specification; >it does not use terms from other vocabularies instead of ones defined in >the vocabulary that could reasonably be used." > >I would rephrase this into: >"A data interchange that is is conformant with a vocabulary: >MUST include data described with terms of the vocabulary in consistence >with the vocabulary's scope and purpose; >MUST respect the semantics of classes and properties as declared in the >vocabulary specification; >SHOULD NOT use terms from other vocabularies in place of the ones defined >in the vocabulary and that could reasonably be used." > >The same with the sentences about vocabulary profile. It currently reads: >"A vocabulary profile is a specification that adds additional constraints >to it. Such additional constraints in a profile may include: >a minimum set of terms that must be used; >classes and properties not covered in the vocabulary; >controlled vocabularies or URI sets as acceptable values for properties." > >We could say: >"A vocabulary profile is a specification that adds additional constraints >to it. A profile: >MAY define the minimum set of terms that must be used; >MAY add classes and properties not covered in the vocabulary; >MAY determine the use of controlled vocabularies or URI sets as acceptable >values for properties." > >I found the list under "5. Vocabulary Discovery Checklist" unbalanced. It >has some verbs (suggesting a checklist) but then suddenly "How to find >vocabularies" and "Where..." appears in the list. I think this should be >harmonized. > >Best regards, >Joćo Paulo > > > > > > > >On 4/12/13, 3:34 PM, "Dave Reynolds" <dave.e.reynolds@gmail.com> wrote: > >>Review comments on [1] as downloaded on 2013-12-04. >> >>These are mostly editorial comments, some are more substantial but none >>fatal. >> >># Overall >> >>The document is in much better shape that it was that last time I looked >>at it, well done! It does seem worth publishing as a working group note. >> >># Document structure >> >>The section numbering is odd. I suspect all the sections are intended to >>be at the same level but currently it comes out with 99% of the document >>as subsections of section 1 then the short "Stability Properties" as >>section 2. >> >># Abstract >> >>Reads oddly, three abstracts in one. Suggest just using the text of the >>later "Purpose of the document section" section. >> >>If you retain the current text then: >> s/conventions through/conventions for/ >> >># Scope >> >>The reference on RDF is given as RDF-SCHEMA which doesn't seem >>appropriate, suggest using RDF Concepts. >> >># 1. Summary of Best practices >> >>This section doesn't really line up that well with the things discussed >>in the document but on balance is probably better to have it than not >>have it, just. >> >>## Box "NOMINATE A PILOT" >> >>Suggest dropping. The linked section doesn't actually talk about >>nominating pilots and that's not always appropriate. People reading this >>document could be at many different stages in linked data publication >>experience. >> >>## Box "SERIALIZATION" >> >>"triplifaction" is not common terminology. >> >>The title and content imply just serialization whereas presumably you >>mean data conversion as well. >> >>Suggest: >> >>""" >>DATA CONVERSION Convert the sources data to a Linked Data >>representation. This will typically mean mapping the source data to a >>set of RDF statements about entities described by the data. These >>statements can then be serialized into a range of RDF serializations >>including Turtle, N-Triples, JSON-LD, (X)HTML with embedded RDFa and >>RDF/XML. >>""" >> >>## BOX DOMAIN >> >>s/on authoritative/on an authoritative/ >> >># 1.3 Vocabulary Discovery Checklist >> >>This section seems weak but I can't think an easy way to improve it. I >>guess just leave it. >> >># 1.4 Using SKOS to Create a Controlled Vocabularly >> >>Out of place, this should come later after at least 1.6 (Vocabulary >>Creation) and probably after 1.7 (Multilingual Vocabularies). >> >>The box "It is a good practice to use SKOS in the following situations:" >>needs work. It mixes up indications when SKOS is appropriate (first 1-2 >>bullets) with advice on how to do it (next 2-3 bullets) with something >>completely odd (last bullet). >> >>Suggest: >> >>""" >>SKOS is appropriate in the following situations: >> >> * There is a need to publish a controlled list of terms or >>taxonomies having a special meaning for the domain. >> * The complexity and formality of an OWL ontology is not appropriate >> (for example the terms are not themselves entities that will be richly >>described). >> >>In creating a SKOS vocabulary bear the following good practice in mind: >> >> * Make a clear distinction between the collections of concepts >>(ConceptScheme) and the different individual concepts. >> * Define, when possible, a different namespace for each >>skos:ConceptScheme. >> * Structure the concepts in the list using properties >>skos:hasTopConcept, skos:broader, skos:narrower. >> * Consider defining a Class to represent all the skos:Concepts in >>your controlled list (this can facilitate declaration of properties that >>will use this list). >> * Provide multilingual labels for the terms. >>""" >> >># 1.6 Vocabulary Creation >> >>The box on versioning policy talks about a best practices section on >>Versioning that no longer exists, suggest dropping the last sentence. >> >># 1.7 Multilingual Vocabularies >> >>s/represents the latest trend in/is an approach to/ >> >>[Not sure judgements like "latest trend" are appropriate here.] >> >>s/ , , /, / >> >>s/morphosintactic/morphosyntactic/ >> >># 1.8 Best Practice for Choosing Entity URIs >> >>The Q answer boxes are hard to read: >> - Some boxes use "Q/Y/N" others use "Q/A" suggest standardizing on the >>former >> - The nesting of questions in the second box in unclear, using braces, >>indenting or something to help people see which Q each N lines up with. >> >># 1.9 URI Design principles >> >>The final section on TAG advice is hard to read for anyone doesn't >>already know all about this, needs some rewrite. >> >>EXAMPLE 2 presumably shouldn't be called an "example", a Note? an advice >>box? Also drop the pre formatting, it doesn't print properly. >> >>Either drop the term http-range-14 or put that up front and explain it. >> >># 1.11 IRIS >> >>s/homegenic/uniformly/ >> >># 1.13 Security and Hosting >> >>This whole section seems very related to the US Government specific >>process. It is not clear to me that it says anything specific enough to >>be useful in a non-US context. >> >>Suggest dropping most of the section and replace by a simpler statement >>to the effect that "that for specific data and organizations there may >>be particular security considerations, so involve appropriate advisors >>early on in the process. Detailed considerations of security issues are >>beyond the scope of this document" >> >>Dave >> >>[1] https://dvcs.w3.org/hg/gld/raw-file/default/bp/index.html >> > >
Received on Thursday, 12 December 2013 13:27:33 UTC