Creating a public dataset

[adunning]

We had a request for a CSV version of our manuscripts list, and it strikes me that we could make a useful public dataset out of this. Any suggestions to improve on the following (thinking of what can export reliably and is fairly normalized)?

• Shelfmark
• SC number
• msName
• date (3–4 columns? for text, notBefore, notAfter, and any precise date given)
• place of origin (ID only? ID+text?)
• head or summary contents if unavailable (as generated online)
• languages (as ISO codes?)
• IDs + text for works in msItem
• IDs + names for former owners
• support material
• dimensions
• second folio
• types from decoNote

@andrew-morrison has suggested creating a separate line for each part in the case of composite manuscripts.

[andrew-morrison]

A lot of the same fields (but not all) were mapped to CIDOC-CRM and FRBRoo ontologies and exported as RDF for the Mapping Manuscript Migrations project. It would probably be possible to formulate a SPARQL query to run against their triple store, although the data is a snapshot from last year. I believe the idea is we'll eventually publish linked data of our own, updated regularly, which would cater for the more advanced researchers. But a simpler CSV export could also be useful for those who just want something they can copy and paste into a spreadsheet.

[tobyburrows]

The MMM project's RDF version of the Bodleian data can be downloaded from the Zenodo repository here: (https://zenodo.org/record/3667486) There's a separate Turtle file for the Bodleian data.