Gordon,

Interesting read.

I would have liked to see more about the idea of creating a shared, useful, framework, into which GLAM data can be pulled. Controlled vocabularies only get a brief mention in section 1.5.6 - and even then only in the context of helping AI systems out.  Institutions are a good example of this need: as the report mentions there is still [shamefully] no freely available resource listing all U.K. cultural heritage institutions, and DCMS themselves find this an impediment to marshalling the very data this report is discussing; equally it will hinder users who want to use institution type or location as a filter in their search for relevant information. However, a CH institutions resource is just one of a number of frameworks which could usefully be created as a 'way in' to object data. The report mentions places and well-known archaeological excavations; there are many more we could create.  Some of these are 'down to us' as a community, since the data is too specialized to be served by the generic disambiguation resources which the trial used. Platforms like WikiData can be used to hold the results, but we will need to populate them with subject-specific data.

While there is undoubtedly much variation in cataloguing practice, there is also plenty of analysis going on, so that the wholesale application of AI text-mining techniques may be less relevant than the report suggests. Disambiguating fielded data of a known type is a different, hopefully easier, challenge than identifying names, places, etc. in free text. This is particularly the case when that field is backed up by a 'termlist' or other control mechanism which encourages consistent recording. Following on from the Linked Art workshop at the V&A last week, I have been exploring the idea of adding AAT URLs to an acquisition methods termlist, and thereby making it easier to export catalogue data as Linked Art format RDF - without having to update any of the object records. In general, I think there is plenty of scope for using structured vocabularies in conjunction with AI analysis.

Linked Data is mentioned as a publication technique, before being dismissed as being too "complicated" (for all but the BM and NG, I'm guessing). This may be fair comment, but I'm disappointed that there is no exploration of the role which [Linked Data] persistent identifiers could play in the data enhancement process which the report describes.  There is much discussion of the successes and failures of AI systems when wrenching unambiguous context from free-text sources, but no consideration of the fact that many of those disambiguated concepts will ultimately be represented by a persistent URL.   Personally I think the whole resulting framework should be a Linked Data thing.

Best wishes,

Richard

[log in to unmask]">

Dear All,

 

I would like draw your attention to our report on mapping digitised collections, which the Department for Digital, Culture, Media and Sport (DCMS) commissioned earlier this year and which has just been published.

 

DCMS asked us and Knowledge Integration to consider, from a technical point of view, how the 2016 Culture White Paper ambition to ‘access particular collections in depth as well as search across all collections’ might be realised.

 

You can download the report here:  

 

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/829225/Mapping_digitised_collections_in_England_-_final_report.pdf

 

In the coming months we will also post some of its key ideas and findings in our blog. The first one is already live. It focuses on why we need to share collections information, and you can read it here:

 

https://collectionstrust.org.uk/blog/mapping-digitised-collections-how-and-why/?utm_source=newsletter&utm_medium=email

 

If you have any comments or thoughts on this important subject I’d love to hear them.

 

Kind Regards,

 

Gordon.

 

--------------------------------------------------------------------

 

Gordon McKenna
Standards Manager
Collections Trust
0207 059 9750

Website | Newsletter | Twitter | LinkedIn | Facebook

Company Registration No: 1300565. Company registered in England and Wales. Registered Charity No: 273984. Registered Office: Collections Trust, Rich Mix, 35-47 Bethnal Green Road, London E1 6LA.

 

---

To unsubscribe from the MCG list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=MCG&A=1

To unsubscribe from the MCG list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=MCG&A=1