archivesnationalesfr / rico-converter

A tool to convert EAC-CPF and EAD 2002 XML files to RDF datasets conforming to Records in Contexts Ontology (RiC-O)

RiC-O converter

A tool to convert EAC-CPF and EAD 2002 XML files to RDF datasets conforming to Records in Contexts Ontology (RiC-O).

This repository includes the converter, some examples files and documentation.

The converter is written mostly using XSLT stylesheets, wrapped in a Java command-line application.

Documentation

The documentation can be found at https://archivesnationalesfr.github.io/rico-converter

References

• Records in Contexts-Conceptual model (RiC-CM) latest public version (dated July 2021): https://www.ica.org/sites/default/files/ric-cm-02_july2021_0.pdf
• Records in Contexts-Ontology (RiC-O) latest official release: https://www.ica.org/standards/RiC/ontology
• Records in Contexts-Ontology (RiC-O) information web site: https://ica-egad.github.io/RiC-O/
• Encoded Archival Context-Corporate Bodies, Persons, Families (EAC-CPF XML schema): https://eac.staatsbibliothek-berlin.de/
• Encoded Archival Description (EAD): https://www.loc.gov/ead/

rico-converter's Issues

Check for empty values when generating literal properties (everywhere)

A query to check for empty literal values in the POC data:

PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
SELECT ?p (COUNT(?s) AS ?count) WHERE {
  ?s ?p ?o
  FILTER(STR(?o) = "")
} GROUP BY ?p

Yields the following results:

1	rdfs:label	"29"^^xsd:integer
2	rico:descriptiveNote	"3"^^xsd:integer
3	rico:textualValue	"14"^^xsd:integer
4	rico:identifier	"6390"^^xsd:integer

Looks like they are rico:identifier on Instantiation such as http://data.archives-nationales.culture.gouv.fr/instantiation/041185-Grandrye_Pierre_0-i1

Generate shortcut properties

Générer systématiquement le triplet de type ‘shortcut’ utilisant la
object property déclarée pour cela dans l’ontologie, dans le cas d’une
relation complexe (impliquant l’instance d’une sous-classe de Relation)
Voir par ex. https://www.ica.org/standards/RiC/ontology#performsOrPerformed
Cette évolution concerne surtout la conversion des notices EAC-CPF. Elle impliquera également de modifier les tests unitaires.

hard-coded h4 in ead2rico.xslt

Thanks for the great work. I noticed that there are many hard-coded html:h4...</html:h4> in French within the ead2rico.xslt. For example, line 121, 127, 159. Is there a workaround for other languages? Many thanks.

Add website structure in French

(use i18n technique with 2 subfolders "fr" / "en")

New version of organization types (remove LegalStatus from the conversion)

EAC to RiC-O : Prendre en compte la nouvelle version du vocabulaire des types d’organismes (supprimer le vocabulaire LegalStatus de la conversion)

Fix hasDocumentaryFormType

Correction dans l’utilisation de hasDocumentaryFormType (à utiliser uniquement pour les classes
Record et Record Part). Utiliser, dans le cas de Record Set, les propriétés similaires aux nouvelles
propriétés qui ont pour range Language (voir plus haut).

Check for proper URL format (encoding, spaces) whe nconverting

Handle cases where <entityId> contains a Wikidata identifier

EAC to RiC-O : Prendre en compte le cas de entityId ayant pour valeur un id Wikidata (use case AN) (similaire BNF et SIRET)

Rename AgentSubordinationRelation to AgentControlRelation (?)

I can't find AgentSubordinationRelation in latest RiC-O.

rico-converter uses it : https://github.com/ArchivesNationalesFR/rico-converter/blob/master/ricoconverter/ricoconverter-convert/src/main/resources/xslt_eac/eac2rico-relations.xslt#L101

Is it to be replaced with AgentControlRelation ? or deleted ?

Migrate documentation to Github pages

Update documentation with latest enhancements

Migrate README.md to the docs subfolder so that it is in the website

Improve <unitDate> processing to generate a single rico:date when @normal contains a period with identical start and end date

Express sequences of Records or Record Sets in a Record Set

We forgot to do so :(, and it is important of course: if we do not do this, we shall lose some information compared to what the EAD sequence of c elements means.
So I think we should do it in the v3 / RiC-O 1.0 version.
We could at least use the new RiC-O rico:directlyPrecedesInSequence object property to connect a Record Resource (whose description is in a EAD c element) to the one that is described by the first following-sibling c element. Same for connecting the Record Resource to the last preceding-sibling c element (using rico:directlyFollowsInSequence).
See also this issue, opened a few days ago about RiC-O: ICA-EGAD/RiC-O#97

Replace hasLegalStatus by hasOrHadCorporateBodyType

hasLegalStatus dans le cas où l’agent est un CorporateBody doit être remplacé par
hasOrHadCorporateBodyType

See also #16

Change ISNI namespace and adjust identifierValid to owl:sameAs

Espace de nom ISNI à modifier (xmlns:isni="https://isni.org/ontology#") et supprimer
isni:identifierValid (remplacer par owl:sameAs)

Wrong version in .bat file

bat file contains 2.0.0-beta and is not synched with the version of the project

Fix includesOrIncluded

Correction à faire dans l’emploi de includesOrIncluded : à remplacer par hasOrHadConstituent
lorsque la cible est un RecordPart (et idem pour relation inverse)

Handle cases where <languageDeclaration> is empty

Prise en compte dans <control> du cas ou <languageDeclaration> est vide (prendre “fr” par
défaut dans la XSLT)

"fr" will be set by default in the XSLT

Implement a global "driver" system to allow for easy overwrite of some rules

A master XSLT that imports the main XSLT, and in which custom rules can be added

Update EAD and EAC example files included in the package

Also, add a few ?

Process values of XML attribute "otherlevel" when level="otherlevel"

See #10

Update language controlled vocabulary file

See comments on #31 :

So we should also modify the VOCABULARY_LANGUAGES param and the eac2rico:URI-Language function in the eac2rico-uris.xslt file + upload the new version of the vocab and delete the previous one. So, do we do this in the same branch or should we rather create a new issue and branch to handle this?

• Take the updated file from vocabulary repo
• Update the reading of the vocabulary to read skos:Concept and not rico:Language

Differenciate between Person and Organization when relating an agent to an activity

différencier pour les relations entre l’agent et une activité, le cas des personnes et le cas des
organismes

Change archdesc URI generation ("top-" before identifier)

URI des ressources RDF correspondant au niveau : la chaîne de caractères ‘top-‘ est à
placer avant l’identifiant de l’instrument de recherche

Rename properties to be conformant to RiC-O 0.2

See the mapping from old to new properties in the attached XML file :

updated-object-properties.zip

Some properties deserve special refactoring, see other issue for this.

Typo in ricoconverter.sh

Hi! I'm testing the converter in a Mac environment and it seems to me that there is a little typo in the "ricoconverter.sh" file.
At line 23 appears a "@" that causes an error while trying to launch a command, since it seems to interfere with the file path, giving e.g. the error:

"ERROR : Could not read file @parameters/convert_eac.properties: java.nio.file.NoSuchFileException: @parameters/convert_eac.properties"

I tried and deleted the "@" at line 23 and the file works correctly. But I'm not an expert at all, so I wanted to check with you the correctness of this change.
Thank you very much!
Lucia

Renaming and refactoring to generate new properties hasContentOfType, hasOrHadLanguage, hasOrHadLegalStatus, hasRecordState

Necessitate a refactoring of the RecordSet.
See also #3

Une attention particulière doit être apportée aux propriétés suivantes : hasContentOfType,
hasOrHadLanguage (qui remplace HasLanguage), hasOrHadLegalStatus (qui remplace
hasLegalStatus), hasRecordState (qui remplace hasRecordResourceState). Le domaine de ces
propriétés a été réduit aux classes Record et Record Part (et Agent pour les propriétés
hasOrHadLanguage et hasOrHadLegalStatus) alors qu’il était autrefois fixé à RecordResource.

And also:

Pour la classe RecordSet, de nouvelles propriétés ont été introduites. Par ex., pour relier un
RecordSet et un Language, on utilise désormais les propriétés suivantes :
hasOrHadAllMembersWithLanguage
hasOrHadSomeMembersWithLanguage
Voir à ce sujet le ticket ICA-EGAD/RiC-O#20
Ce qui conduira dans RiC-O Converter à réécrire les instructions correspondantes.

Shortcut properties for mandates and activities

See comments #5 (comment) and #17 (comment) : we need to set a rdf:nodeID + refer to it in the shortcut property.

Handle the case where entityId has a localType attribute with values "numéro RNA", "numéro SIREN" or "numéro SIRET"

These are quite recently added values for this attribute in the ANF information system.

For now, RiC-O Converter only takes into account, for this localType attribute, the "ISNI" value, + the "Wikidata" one (see #15).
It also handles the following case:

<entityId>SIRET 501 524 029 00086</entityId>

not more.