Property talk:P3628

I feel this property was created prematurely, while we were still discussing what is needed. What's the best way forward? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 10:16, 19 February 2017 (UTC)Reply

From the discussion it seemed that everybody (perhaps apart from @Andrew Gray: who hadn't re-commented) appeared happy to use the property as proposed.

My revised view, as given there, was:

"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 (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."

We can discuss what the appropriate form for those full bibliographic references might be; my view, as stated, is that such bibliographical references, if/when they get added, would need to be at least at the book level, if they are to convey eg author and date information. But as per DNB00, if there is a transcription effort going on at Wikisource, then there could be a case for a bibliographic item at article level. Jheald (talk) 10:33, 19 February 2017 (UTC)Reply

While it was indeed a helpful discussion, no bones broken, really. I don't think this VCH property is superfluous to needs: anyone (for example) working to create parish items here for English places, where enWP has an article that covers both the place and the associated parish, is going to find it pretty useful. I'm all for process wonkery in its right place, but (a) Andrew didn't oppose, which would have been a brake on the process, and (b) Wikidata is yet young, so the calibration of process closures is not going to be as accurate as it might be. Charles Matthews (talk) 11:00, 19 February 2017 (UTC)Reply

I'm a bit ambivalent - I think it's a bit of an odd approach to go down this route when there isn't a clear 1-1 correspondence between these "identifiers" and the sort of subjects we want to use, and we might want to rethink it later on, but on the other hand, no harm done as Charles says and it's not like this won't be of some use as is. Perhaps we could restrict this to parishes/villages/wards for the time being, rather than (eg) placing this on items for buildings, charities, schools, families? That seems to be the clearest way our entries map to theirs. Andrew Gray (talk) 11:20, 19 February 2017 (UTC)Reply

@Andrew Gray: Why should it be a problem to map eg St Bartholomew's Church, Edgbaston (Q7592600) to a particular paragraph, eg warks/vol7/pp361-379#p5 ? Jheald (talk) 12:11, 19 February 2017 (UTC)Reply

Interesting, I'd forgotten you could resolve it on such a low level! If the constraints can make sense of this, then sure. Andrew Gray (talk) 12:14, 19 February 2017 (UTC)Reply

(ec) I agree that it would be a bad thing to map without the anchor to a particular paragraph. But such cases can probably be picked up for refinement on an ongoing basis by their tripping up an exclusivity constraint requirement, or similar query.

I don't think we can enforce that individual buildings must link to a particular paragraph, because eg cathedrals might get a whole chapter. Jheald (talk) 12:17, 19 February 2017 (UTC)Reply

@Charles Matthews, Andrew Gray, Pigsonthewing: What's the best thing to do with entries like the Borough of Lewes, which stretches over 3 chapters/webpages in Sussex vol 7 ?

Or Newcastle under Lyme, which takes up almost quarter of a book in Staffordshire vol 8 ?

If we just link to the first page, then the reader may not realise how much more there is, and won't see the navigation / table of contents.

On the other hand, if we link to the table of contents, then multiple items will have the same link. And unfortunately there don't seem to be any anchors on the table-of-contents page, to jump it to the right place.

I suppose we could just create a fake anchor, eg http://www.british-history.ac.uk/vch/staffs/vol8#p1-75 and let the reader work it out? Jheald (talk) 23:11, 1 March 2017 (UTC)Reply

First data now going in, 1975 values that were linked from pages at A Vision of Britain through Time (Q28925900).

This initial set are for where VoB linked to only one VCH entry (often it links to many, as per the section above); and where the name of the item matches the title of the VCH article. I also intend to create bibliographic-type references as well, once we can see how these initial P3628s look. Coverage in this inital set is limited to only a subset of volumes (93 distinct volumes in all) - presumably those that were online at BHO at the time when the GB HGIS team made their extraction. Jheald (talk) 14:13, 11 March 2017 (UTC)Reply

Query results look reasonable:

• Uniqueness search currently producing two pairs: Seamer (Q2767197)/Seamer (Q7440719), Brompton (Q2300765)/Brompton (Q4973754) -- both examples of similarly-named villages located quite near each other, probably indicating at least one has an incorrect Vision of Britain place ID (P3616) match.

Now fixed. In both cases the relevant one was the one near Scarborough. Both the VoB assignments were right, so this misidentification stemmed from the VoB site itself. Jheald (talk) 17:45, 11 March 2017 (UTC)Reply

• Multi-value search returning nothing, by construction, as intended.
• Map display shows very clearly the patchiness of this first extraction -- areas either covered very heavily, or not at all.
• List of classes looks pretty solid, albeit a few oddities at the bottom. These are probably from questionable identifications made for Vision of Britain place ID (P3616) -- if this is the case, please check whether other statements may be wrongly on the item too, and move them too to the item they should be on.
• List of co-properties shows up fairly clearly where this first tranche has come from - ie Vision of Britain places that has been matched by co-ordinates. No great horrors or surprises in the rest (I'd say), though some obviously could use a little more populating.

Jheald (talk) 16:48, 11 March 2017 (UTC)Reply

Note that we need to use the Toolforge URL formatter script, rather than just the default, because the default formatter changes # into %23, which breaks the URLs. (eg broken vs correct).

Sending the IDs to the Toolforge script works around this. Jheald (talk) 08:54, 10 June 2022 (UTC)Reply