dm2e / dm2e-mappings Goto Github PK

Is there any reason, why a dc:subject is not repeatable?

Please correct that in the specs and validation tool. I can see no advantage in this decision.

The shownAt link seems broken as it lands on a [Empty Page], while the AnnotatableVersion does contain text.

See for example this one:

http://data.dm2e.eu/data/html/resourcemap/mpiwg/harriot/MPIWG_01QU84RT_00793/1393267718023

http://data.dm2e.eu/data/rdf/resourcemap/uib/wab/Ms-114/Ms-114%2C145r%5B2%5Det145r%5B3%5D/1391118905391
here the format is a resource:
dc:format http://onto.dm2e.eu/schemas/dm2e/1.1/mime-types/text/html-named-content .
, while in other datasets a label was used (html-pundit-content). What is the correct value?

Currently the validator allows arbitrary URI resources as object in triples

?cho dc:type ?dc_type

?dc_type must be one of the enumerated URI values listed in 3.7

The data at:
http://data.dm2e.eu/data/html/dataset/ub-ffm/sammlungen

is using now an "UNOFFICIAL" namespace.
Also datasets are empty.

http://data.dm2e.eu/data/html/dataset/uib/wab/1393434327660

Are there images somewhere?
I understand CHOs are transcriptions of manuscript pages, did you also provide facsimiles of such pages?

Should the validation rely on a strict closed world assumption and consider properties that are not defined by the DM2E data model as an error or a mere warning?

Right now, I collect all DatatypeProperties and ObjectProperties of the resp. OWL file and iterate through all the properties in the data to check. If a property is not in this whitelist, a WARNING is given.

But in most cases, these are actually errors, e.g.

<http://data.dm2e.eu/data/item/mpiwg/harriot/MPIWG:01QU84RT_982>:.
    [WARNING]   UNKNOWN_PROPERTY    http://purl.org/dc/terms/description
    [WARNING]   UNKNOWN_PROPERTY    http://purl.org/ontology/bibo/number/
    [WARNING]   UNKNOWN_PROPERTY    http://purl.org/dc/terms/title

1 and 3 should be dc elements not dcterms, 2 is a typo.

So I suggest making unknown properties errors.

I cannot consume resources like

http://data.dm2e.eu/data/agent/ub-ffm/sammlungen/Universit%C3%A4tsbibliothek_JCS_Frankfurt_am_Main

They take ages to load as the contains a lot of triples.

However I need a way of getting at least the label of the content provider.
Would it be possible to put the relative triple in the representation of the CHO? so I do not have to follow this link?

What other alternatives?

http://data.dm2e.eu/data/html/dataset/uib-ffm/sammlungen/1390576653139
here the contained CHOs are Pages, there is no Book level: http://data.dm2e.eu/data/html/resourcemap/uib-ffm/sammlungen/urn:nbn🇩🇪hebis:30:2-11683-phys1970992/1390576653139
Note that the rdf:type used is http://purl.org/spar/fabio/#Page and not dm2e:1.0/Page as in other datasets.

When I check my data with the latest validation-tool with -version 1.1_Rev1.4, the following error appears:

http://data.dm2e.eu/data/agent/bbaw/authority_gnd/1018099549:
[ERROR] has unknown property http://www.w3.org/2002/07/owl#sameAs

But according the spec, it is optional (p. 53)

Wrong namespace?

http://data.dm2e.eu/data/html/dataset/onb/codi/1391095292351, http://data.dm2e.eu/data/html/dataset/onb/codices/1391337499559,
this seems empty

Here I collect issues I encounter while working with ABO mappings and the validator 1.1_Rev1.5-DRAFT

1. foaf:Organization not allowed as dc:subject --> plz fix
   [ERROR] INVALID_OBJECT_PROPERTY_RANGE http://data.dm2e.eu/data/agent/onb/authority_gnd/4029989-2

2. foaf:Person not allowed as dc:subject --> plz fix
   [ERROR] INVALID_OBJECT_PROPERTY_RANGE http://data.dm2e.eu/data/agent/onb/authority_gnd/118838989

3. bibo:Series not allowed as dc:type --> FATAL, plz fix
   [FATAL] INVALID_DC_TYPE http://purl.org/ontology/bibo/Series

4. edm:Place as dc:subject throws error, plz fix
   [ERROR] INVALID_OBJECT_PROPERTY_RANGE http://data.dm2e.eu/data/place/onb/authority_gnd/4074987-3

----> In the meantime I suspect that the ranges for dc:subject are generally flawed -> OWL file?

Example:

http://data.dm2e.eu/data/rdf/resourcemap/mpiwg/harriot/MPIWG_01QU84RT_00817/1393267718023

there is not triple specifying the format.

What should the format be? text/html?

http://data.dm2e.eu/data/html/dataset/gei/gei-digital/1391717717253

.. or I'm looking at the wrong version? please delete old versions!

[ERROR] UNKNOWN_PROPERTY http://rdvocab.info/ElementsGr2/otherDesignationAssociatedWithTheCorporateBody

This property is already part of Rev1.3

http://data.dm2e.eu/data/html/dataset/ub-ffm/sammlungen/20140301004538912

Everything that should show up in the search results should have a title.

According to the EDM v.5.2.4 (http://pro.europeana.eu/documents/900548/0d0f6ec3-1905-4c4f-96c8-1d817c03123c) and the Europeana Portal Image Policy (http://pro.europeana.eu/documents/900548/960640/Europeana+Portal+Image+Policy), the WebResource referenced by edm:object in an ore:Aggregation should be:

• the URL to an image representation of the CHO in the highest resolution available on the provider's web site
• in an image format supported by the image processing library ImageMagick
• at least 200px wide

Since this is the property that both Europeana Portal and DM2E will use for display in the respective search interfaces, it would simplify the thumbnail generation process if we knew what MIME type such a WebResource has.

Therefore I propose to make dc:format mandatory for WebResources that are the related to an ore:Aggregation using the edm:object property.

The following MIME types should be supported:

• image/png
• image/jpeg
• image/gif
• image/tiff
• application/pdf

PDF may only be used if the first page of the PDF is the representation of the CHO.

Resource like the following

http://data.dm2e.eu/data/html/resourcemap/uber/dingler/page_pj022_pb150/1393515961557

have an Annotatable Version that is an HTML page.
Is contains both text (trascription) and images (facsimile).
It would be VERY IMPORTANT to SHOW THEM IN THE DEMO.

However:
Images are enclosed in a viewer and text is somewhere in the HTML DOM.
Does this respect the specifications we agreed? I would like to have links to Images and texts, if possible separately. so I can aggregate them in Feed and annotate them.

Question: what is the why of connecting a transcription to its facsimile? Is the model supporting this?

Representing dates like resources is ok.
Example:
http://data.dm2e.eu/data/timespan/onb/abo/1763-01-01T000000UG_1763-12-31T235959UG

But consuming this is very heavy as I need to get a full RDF resources with a lot of links that I do not need.
Would it be possible to include the date also as Literal in the RDF description of a e.g. persons?

here the ? indicates some broken namespace, I guess:

http://data.dm2e.eu/data/html/agent/onb/authority_gnd/100952798

Please change [WARNING]s, which are based on recommendations, to [INFO].

That's much more friendly.

(Thanks @kba! Finished. And announce updates.)

First: there are no problems, there are only challenges!

What does it mean?

In the case of dta, edm:isShownAt, edm:isShownBy and edm:object point to different resources, according to the spec.

If there is a problem, the spec is too complex.

[FATAL]  is missing required property  (Condition: Annotatable WebResource)

Not every WebResource is annotatable, e.g. the WebResource for a book points to a landing page. I am not sure, this validation is too strict.

I fixed it in my code anyway.

http://data.dm2e.eu/data/dataset/mpiwg/rare

Example:

http://data.dm2e.eu/data/html/resourcemap/uber/dingler/page_pj022_pb150/1393515961557

there is no dc:format for the annotatable resource.

http://data.dm2e.eu/data/rdf/resourcemap/uib-ffm/sammlungen/urn:nbn🇩🇪hebis:30:2-11683-phys1971057/1390576653139
there is no dc:format scpecified for the Annotatable object!

CHOs are single pages? No CHO (boo)k exists that contains them.
It is correct?
How should I treat them?

I am trying to run

java -jar dm2e-validate.jar -version "1.1_Rev1.4" -terse test.rdf

on a file successfully validated in http://www.w3.org/RDF/Validator/rdfval

and I only get the following error message:

! Jena croaked on file /media/sf_devel/test.rdf. Are you sure it is 'RDF/XML'. http://RELATIVE_URL/ Code: 11/LOWERCASE_PREFERRED in HOST: lowercase is preferred in this component

Please find the content of the file here http://wiki.dm2e.eu/wiki/images/5/51/Validate_Test.zip

dm2e-validation.jar NOTICEs:

has untyped literal for property .

The values in the file are 2014-01-09T23:16:06Z, which are valid xsd:dateTime.

DM2E-spec lists edm:TimeSpan, xsd:dateTime, rdf:Literal as possible ranges for dcterms.

Why is it NOTICEd? Or what is a better notation?

bibo:number is mandatory and has datatype xsd:int.

When mapping encounters a page without a page number what should data providers set as a value here. They must set some value and it must be a number.

Probably -1 is a good idea though this will break the naive find-the-first-page algorithm.

Otherwise bibo:number should not be mandatory.

Link: http://data.dm2e.eu/data/html/list
Question marks in place of URL prefixes usually denotes some mulfunctioning (e.g. ?:hasCollection)

Link: http://data-worker.dm2e.hu-berlin.de/data/html/resourcemap/bbaw/dta/16157/1391002599515
a namespace is broken : ?:author (http://purl.org/spar/pro/author)

Example:

http://data.dm2e.eu/data/rdf/resourcemap/mpiwg/harriot/MPIWG_KECMZ5BH_00003/1393267718023

In this RDF I see no aggregatedCHO property connecting the Item (CHO) with the Aggregation. This is mandatory!

If I encounter a CHO with dm2e:displayLevel "true"^^xsd:boolean whose corresponding ore:Aggregation has neither edm:isShownBy, edm:object nor dm2e:hasAnnotatableVersion, what should be displayed as

• the annotatable version
• the thumbnail?

Should I skip those links altogether?

Should I select the first page for annotation/as thumbnail? How do I find the first page?

Example:
http://data.dm2e.eu/data/html/resourcemap/onb/abo/%2BZ103504606/1391106413338

there is no edm:object triple. This is needed to provide a preview of the book.
Isn't it mandatory in the mapping?

http://data.dm2e.eu/data/html/dataset/mpiwg/rare/1391777127562 resources are broken. e.g. http://data.dm2e.eu/data/html/resourcemap/mpiwg/rareBooks/MPIWG_00N0SGXE/-1

Example:
http://data.dm2e.eu/data/html/resourcemap/uib-ffm/sammlungen/urn:nbn🇩🇪hebis:30:2-11683-phys1970999/1390576653139

http://purl.org/spar/fabio/#Page is wrong, you should use dm2e:Page!

• bibo:number
• dm2e:levelOfHierarchy
• bibo:numPages

dm2e:displayLevel is in the domain or ore:Aggregation; dm2e:levelOfHierarchy should be there as well

I cannot get what is the correct version of the dataset, some are empty some have messy data.

Please delete the old datasets if possible! Or clearly give a name to the version last version.

Example:
http://data.dm2e.eu/data/html/resourcemap/bbaw/dta/16168_f0004/1391002599515

in this RDF the URL http://purl.org/ontology/bibonumber seems wrong.
I guess the correct one is http://purl.org/ontology/bibo/number

http://data.dm2e.eu/data/html/dataset/uber/dingler/1392199702509

In "3.7 edm:PhysicalThing: Subclasses" there is no Class defined for an Issue of a serial work. This should be bibo:issue I think.

http://141.20.126.232/data/html/resourcemap/onb/abo/%2BZ101315405/1391106413338
is this correct? : http://141.20.126.232/data/... shouldn't it be a domain name?
There is not author triple. Is it correct?
The rdf:type is Book but it does not contain pages. Is it correct? What should I annotate then?

http://data.dm2e.eu/data/html/resourcemap/uib/wab/Ms-114/1391118905391
there is no rdf:type that allows me to assume this is a Book and contains pages (perhaps Book was used as a type in other datasets).

Just came across such a case:

  <dm2e:incipit>Prospero prudente constate felyce ... [Prolog]</dm2e:incipit>
  <dm2e:incipit>Entonçes se apareja la cosa ...</dm2e:incipit>
  <dm2e:explicit>... Fable la obra que el dezir se çierra. [Prolog]</dm2e:explicit>
  <dm2e:explicit>... ny demandarles cosas non acostumbradas.</dm2e:explicit>