Rob Smith @natlibscot.bsky.social talks about enriching the National Bibliography of Scotland dataset with Linked Open Data from authority records using Python and APIs
#MDGSBites

@librarywales.bsky.social have spent a lot of time on reconciling entities from their dataset with Wikidata.
They are also exploring potential applications for LLMs in supporting entity recognition in digitised text, which can then be reconciled & matched with the SNARC wikibase #MDGSBites

Joining our friends at @mdgscotland.bsky.social for a #MDGSBites webinar on Linked Open Data in practice. Looking forward to hearing two different perspectives of #LinkedOpenData from Rob Smith @natlibscot.bsky.social and Jason Evans @librarywales.bsky.social
#metadata #PIDs #URIs

Wikidata is open to all, a community built dataset, anyone can edit it.
A bridge is needed between Wikidata & an archival database via a custom Wikibase to enable more control over who edits the data
Semantic Name Authority Repository Cymru
snarc-llgc.wikibase.cloud/wiki/Main_Page
#MDGSBites

Crowdsourcing has also been used to get humans to help with QA of matches - 48,819 name authorities have been matched across the archival data and wikidata.
#MDGSBites

Quality Assessment - verification of the matches between @natlibscot.bsky.social data and @isni-id.bsky.social data.
Auto-enrichment - sending the data back to Alma via an API to enrich the bib records.
#MDGSBites

In development is SNARC GPT an LLM Chatbot that can interrogate the SNARC Wikibase in English or Welsh to find archival, library and digital resources across the collections of @librarywales.bsky.social #MDGSBites

SPARQL queries are used to interrogate the Wikibase and open data sources across @wikidatacommunity.bsky.social
This enables new visualisations of the collections and associated entities such as People, Places, Events etc.
#MDGSBites

This process enabled enrichment of over 3000 records with URIs - and it helped identify Scottish writers/creators with no presence in the LC Name Authority File, ISNI, and/or Wikidata.
#MDGSBites

@librarywales.bsky.social are working to build connections between library, archival, and art collections via Linked Open Data - mostly using OpenRefine to reconcile and match names and dates across archival datasets and Wikidata
#MDGSBites

Next steps will be to look beyond National Bibliography of Sotland to other @natlibscot.bsky.social collections - to enrich other authority controlled fields Names of persons and corporate bodies and possibly also subject headings.
#MDGSBites

The project focussed on embedding ISNI URIs and Wiki URLs into the metadata for Name Headings.
Rob used a Python script to query the ISNI database to find matches based on name and date data.
#MDGSBites

Wikidata enables much richer data to be recorded about a Person via statements called triples (subject-object-predicate) - this enables links to be built between any given Person and other related entities: Persons, Places, Concepts, Subjects, Events....
#MDGSBites

Developments in LOD display in Alma/Primo have enabled these URIs to be used in discovery, such as via the display of 'Info Cards' for persons - leveraging biographical info in LCNAF, ISNI & Wikidata, so it can be displayed alongside associated works
#MDGSBites

Jason from @librarywales.bsky.social talks about using Wikibase for authority control in archival collections.
Authority records can be very sparse - they are more human- than machine-readable. They identify a Person with a Name, but often that's it! They also operate largely in silos.
#MDGSBItes

In Year 2, further Python queries were used to identify NBS records that lacked LOD - and then to query the ISNI database to extract ISNI and Wiki URIs with an associated 'confidence score' and reason for match to support the quality assessment process.
#MDGSBites

They also connect with external datasets from other agencies like cadw.gov.wales
SNARC wikibase helps to manage different relationships between authority records and datasets to link content between different data silos, such as between the archival catalogue and the library catalogue #MDGSBites