Wikidata:Property proposal/British History Online VCH ID

British History Online VCH ID edit

Originally proposed at Wikidata:Property proposal/Authority control

Done: British History Online VCH ID (P3628) (Talk and documentation)

Description	identifier of a place, in the British History Online digitisation of the Victoria County History
Data type	External identifier
Domain	places in Great Britain
Example	Birmingham (Q2256) → warks/vol7/pp1-3
Source	County pages, like http://www.british-history.ac.uk/vch/warks
Formatter URL	http://www.british-history.ac.uk/vch/$1

Motivation

The Victoria County History, started in 1899, is an ongoing, encyclopaedic history of the historic counties of England, coordinated by the Institute of Historical Research in the University of London. en.Wikipedia already has over 2,300 links to pages to the site. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:46, 4 February 2017 (UTC)[reply]

Discussion

Hmm - I don't think this is the best way to do it. Giving a page reference as an identifier doesn't seem quite right to me (and in some cases it's unduly precise - you could as easily argue that Birmingham is vch/warks/vol7, for example)

My preference would be to create items for the VCH volumes/sections and crosslink with described by source (P1343) from the places like we do with DNB entries. This would also allow the same entry to be crosslinked as a source for, say, the manor house and the church and the parish/village as a whole for somewhere like Castle Bromwich, whereas you'd have a bit of trouble doing this using identifiers; IME many of the WP links are for a single building in this way. Andrew Gray (talk) 14:47, 4 February 2017 (UTC)[reply]

Support. The suggestion seems reasonable enough, given that BHO have developed identifiers in this fashion. described by source (P1343) is still somewhat mired in uncertainty, when it comes to subdivisions of a work, and a "manual of style" type discussion is probably going to be needed to clarify the business. Charles Matthews (talk) 17:32, 4 February 2017 (UTC)[reply]
Comment Note that the Vision of Britain site, subject of another current pending proposal, has done an excellent job of cross-referencing itself to the VCH. We should ask Humphrey Southall (previously a definite friend of Wikipedia) if it would be okay with him to scrape the cross-references wholesale. VoB also cross-references itself extensively to Wikipedia, so both links could be grabbed at the same time. Jheald (talk) 00:22, 7 February 2017 (UTC)[reply]
Support. Jheald (talk) 11:52, 15 February 2017 (UTC) Perhaps not Jheald (talk) 10:36, 8 February 2017 (UTC) Support. Jheald (talk) 00:22, 7 February 2017 (UTC)[reply]
Latest view After more consideration, I think the property is a good idea after all. There's no particular reason why this property shouldn't co-exist with the more detailed bibliographic referencing described below. In fact, for practical use, the two complement each other quite well. This is a nice quick lightweight property that someone can add, and a link will show up straight away in the 'External Identifiers' section. Then, perhaps at a later date, someone else can come along and harvest all those links, do some scraping and some scripting, and generate described by source (P1343) statements with all the right qualifiers, to add with Quick Statements. So this property and described by source (P1343) would complement each other nicely, and it would be good to have both. Jheald (talk) 11:52, 15 February 2017 (UTC)[reply]

Earlier view I think Andrew is right, and described by source (P1343) is the way to go, with the qualifier reference URL (P854) used to create a clickable link on the item (as recommended on the documentation page for P1343) to a particular web-page.
As to the right value for described by source (P1343), one could have P1343 -> Victoria County History (Q7926668); but my preference would be to create an item for each volume in the series, and have P1343 point to that, because it is associated with each volume that we have distinctive common bibliographic information needed to build a citation (ie date/editor/publisher etc).< br />In turn, volumes should include the properties part of (P361) and part of the series (P179), so that "A History of the County of Warwick: Volume 7, the City of Birmingham" would be marked part of (P361) "A History of the County of Warwick" and both with part of the series (P179) Victoria County History (Q7926668). This structure should make it easy to query for eg 'all VCH references for Grade I listed buildings in the West Midlands', or whatever.
An advantage of this scheme (rather than an identifier) is that it plays well with any references we may import with citations to the dead-tree volumes, which can sit naturally in it using the page(s) (P304) and section, verse, paragraph, or clause (P958) qualifiers until somebody looks up the URLs for them; in the other direction the page(s) (P304) and section, verse, paragraph, or clause (P958) qualifiers desirable to be able to recreate a full bibliographic citation for each reference can be easily obtained from the recommended citation slug at the top of each URL.
My only slight concern is whether this could easily be accessed by eg a {{VCH}} template for the external links section of a wiki-page, with alternate mode(s) to generate fuller bibliographic citations for inline use. But presumably it is easy enough in Lua to step through all the described by source (P1343) statements on an item and render only the ones with part of the series (P179) Victoria County History (Q7926668). Jheald (talk) 10:36, 8 February 2017 (UTC)[reply]
@Andrew Gray, Jheald: I'm in two minds, and intrigued by this suggestion. Could one of you do this for an item, please. so we have an example to consider? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:46, 8 February 2017 (UTC)[reply]
Hi @Pigsonthewing: I've made a start with Birmingham (Q2256) and St Bartholomew's Church, Edgbaston (Q7592600) now referencing Victoria History of the County of Warwick: Volume 7, the City of Birmingham (Q28733155) in series Victoria History of the County of Warwick (Q28733173) and part of Victoria County History (Q7926668). No doubt @Andrew Gray: and @Charles Matthews:, knowing a lot more than me about items for books and book series (these were my first, I think), and what to name things and stuff, will be able to find a raft of style points that could be improved. But these give some basic first concrete items we can play with and think about.

It may well be that there are tools that can automatically generate an item from a book's OCLC number, which would speed the set-up process. I don't know what the state of the art is in generating a citation from a described by source (P1343), but it might form the basis for a wikidata-driven {{VCH}} template. Jheald (talk) 21:24, 8 February 2017 (UTC)[reply]
Occam's razor suggests not relying so much on items for individual volumes such as Victoria History of the County of Warwick: Volume 7, the City of Birmingham (Q28733155). Another issue of a general kind is "seeing round corners" (here seeing into the "part of" statement), and how it is possible within queries. It must be easier for SPARQL to fish information out of qualifiers: I'd be glad to be proved wrong, but my current understanding is that it wouldn't be superficial to collate "referenced to the VCH" if the style on St Bartholomew's Church, Edgbaston (Q7592600) were adopted.
section, verse, paragraph, or clause (P958) is helpful in doing the qualifier style for stated in (P248) references - but is of string type. Absent (as far as I know) a similar general property to build up references to works with parts ... I'm feeling that a definitive discussion is not yet to be had. Referencing to the DNB is certainly in a mess, at present. Charles Matthews (talk) 15:11, 9 February 2017 (UTC)[reply]
Thanks, @Charles Matthews:. So here are a couple of SPARQL queries:
tinyurl.com/hblcswu for every item described by source (P1343) a part of the VCH.

tinyurl.com/hs2d38g for every item with a statement referenced as stated in (P248) it. (currently returns nothing)

(One might also replace wdt:P361? with (wdt:P361|wdt:P179)*, to include in any combination of part of (P361) and part of the series (P179) statements below the main Victoria County History (Q7926668) object, rather than trusting that everything will be explicitly part of (P361) it.)

so for this sort of search the "part of" doesn't create much difficulty for SPARQL; though of course it's the human who may have to "see round the corner" to recognise that this is how things have been set up. (One reason that I labelled the book items "Victoria History of...", rather than "A History of...", the actual title held in title (P1476) for the items).

Compared to a VCH property that pointed to eg "warks/vol7/pp361-379#p5", that could do the first job (mostly) -- though it's worth noting that there is a lot of the VCH that is not available at British History Online, eg for Warwickshire vol 1 is not yet available at BHO, though it is at the Internet Archive.

For the second job (a reference on a statement), I suppose you could end up with any one, two, or three in parallel of stated in (P248) Victoria County History (Q7926668); reference URL (P854) <url>; or the new VCH property used (somewhat unusually for an external identifier) as a referencing property. In each case, it would be quite hard to reconstruct a dead-tree reference from the data.

I am sensitive to Charles's point about Occam's razor -- I do worry that sometimes it seems the ultimate destination of the citation community here sometimes appears to tend towards reproducing ever more of OCLC, PubMed, etc, etc. here on Wikidata. But on the other hand, we already have Wiltshire Victoria County History (Q8023449), Gloucestershire Victoria County History (Q5572118) and Somerset Victoria County History (Q7559911) reflecting articles on en-wiki, so a search for things cited to the VCH should probably already include things cited to things are part of (P361) or in the part of the series (P179) of the VCH as a whole; and there's a whole category tree under c:Category:Victoria_County_History on Commons, so more items may be coming anyway.

If we want to be able to create full dead-tree citations, then we need to have somewhere to put the information to make them (eg Editor/Title/Date). Of course that may not be a requirement; or it may not be a requirement to store the makings for such citations here centrally on Wikidata, rather in existing templates on individual Wikipedia articles.

Which still leaves us back with the question, is a VCH identifier property useful? Not sure. Jheald (talk) 14:50, 10 February 2017 (UTC)[reply]
Tracked in Phabricator
Task T157811
BTW, I tried to see whether any existing reference URL (P854) pointed to 'www.british-history.ac.uk/vch/', but it seems the total number of references with P854 is now so great, it is now impossible to search them for anything without the SPARQL query timing out. (I'm guessing the underlying problem might be the cast of type from URI -> string, rather than being able to string-search URIs directly; but we shall see). I filed an issue for it. Jheald (talk) 17:10, 10 February 2017 (UTC)[reply]
Support. YULdigitalpreservation (talk) 14:22, 7 February 2017 (UTC)[reply]

@YULdigitalpreservation, Jheald, Charles Matthews, Andrew Gray, Pigsonthewing: Done ChristianKl (talk) 20:32, 15 February 2017 (UTC)[reply]