Wikidata talk:WP EMEW

Home

Discuss

Tasks

Items

Properties

Query examples

Tools & Tips

SpBot archives all sections tagged with {{Section resolved|1=~~~~}} after 7 days.

Ask questions or share ideas about this project here!

Please sign your comments with four tildes (~~~~).

Locations other than as points

Main page refers to objective being places (as points). Does wikidata (and therefore this WikiProject) also give IDs to vectors, curves and polygons?

As I understand Stephen Gadd's intention, EMEW knowledge objects will include vectors, curves and polygons, not just point IDs
 – The preceding unsigned comment was added by ColinStuartGreenstreet (talk • contribs) at 23:28, 18 February 2021‎ (UTC).

@ColinStuartGreenstreet: Wikidata items can be connected via geoshape (P3896) to a shapefile stored as a geoJSON on Commons. Shapefiles can be displayed as a blob by queries, but otherwise they're a black box to WDQS. There is an extended standard, called GeoSPARQL which includes functions to eg test whether a point is in a polygon, or to return the overlap between two polygons. ISTR that the Ordnance Survey has a server running GeoSPARQL on its open data, that the public can run queries on. But it's not something WDQS can do.

Query https://w.wiki/33wV gives a list of items identified as having country (P17) = United Kingdom (Q145) -- but note that there are only 464. The reasons for such a small number I am not entirely sure about -- we might ping Simon Cobb @Sic19: to see if he has thoughts, as Simon hand-crafted and uploaded a fair number of the ones that are there.

(BTW Simon some of the file names seem to be broken on WDQS, per that query, even though right on the wikidata item itself -- eg for Blackburn with Darwen (Q880782) the query says the shapefile is c:Data:Blackburn+with+Darwen.map which fails, rather than c:Data:Blackburn with Darwen.map which works. I think the WDQS information loader may have gone through a phase of doing this, but I thought that had all been caught and sorted ages ago? Evidently not.)

Anyway, as to why there are so few shapefiles? I look forward to User:Sic19's takes, but some possible first thoughts:

(1) shapefiles take a lot more work (and time) to research, prepare, and upload than just a coordinate-pair, and people just haven't had it as a priority;

(2) co-ordinates were there for the taking from the Wikipedias; shapefiles weren't;

(3) because co-ordinates are more queryable, and we have had map-link portals for them for longer, absence of coordinates is more annoying;

(4) there's quite good tech now to (often) be able to pull shapefile for a particular Q-id from OSM, eg to display on a Commons infobox (eg the map on Commons c:Category:London), or to access in a hybrid query (see eg recent hits for 'OSM/Wikidata query' and similar in recent Wikidata newsletters: [1]; so it's less of a problem if they aren't locally here. And after all, it is what OSM do.

So perhaps that's why there are so few shapefiles.

As for linear features, eg River Thames (Q19686) -- see Q19686#P625 or the A1 road (Q279859), we're not very good at representing them. The river has two point coordinate-pairs, one for each end. The A1 only has one, though it does come with a rough jpg raster map, and extensive set of primary destinations (P1302) that could be plotted up on a map. Similarly the course of a river could perhaps be traced by plotting the locations of the bridges over it; and here's a map https://w.wiki/F9m of the Antonine Wall (Q210957) built from the coords of westernmost and easternmost points of listed sections of it . And I suppose Wikidata could store a long enough series of point-coordinates of almost anything. But it's not really native to it, and not a direction anyone's really gone in (as far as I aware).

Finally re your question, it's worth noting that wikidata gives IDs to things -- that then might have points, vectors, curves, polygons associated with them. (Or, as seems likely from the above, not). It's maybe a difference between wikidata and a GIS -- for a GIS the trace on the map may sometimes be primary, with what the trace represents a secondary thing -- a label, or some other decoration, attached to the trace. On wikidata it's the thing that is primary, the focus for all the statements that can be made about it (which might include a co-ordinate pair, or more than one, or a shape - or no geographic information at all). That's the upside, with SPARQL as an amazing language to query those relationships. The downside is that, beyond the rudimentary map-plotting abilities of WDQS, wikidata doesn't have the capabilities for representing points, vectors, curves and polygons, nor the plug-ins for doing interesting things with them, that are bread-and-butter for a GIS like the one behind CarterGraph has. But that's one reason why it's quite exciting to see what you can do, if you put the two together! Jheald (talk) 18:00, 4 March 2021 (UTC)

Ontologies for commodities

The Viae Regiae project (and MarineLives) will be looking at (and in the case of MarineLives already looks at) the transportation patterns of different classes of commodity. MarineLives has done extensive work categorising Early Modern commodities. The MarineLives categorisation should be compared with publicly available schema for classifying Early Modern commodities. The Wikiproject Early Modern England and Wales project should look at commodity classifications it already uses, and also industrial sector classifications it aleady uses (the SIC system is a well known modern one). Economic historians have developed their own ontologies to classify economic activities by sector, for example HISCO, which are flawed, but are worth reviewing.  – The preceding unsigned comment was added by ColinStuartGreenstreet (talk • contribs) at 23:42, 18 February 2021‎ (UTC).

@Jneubert: Can I ping you on this? You've done some work on aligning wikidata with economic thesauruses / controlled vocabularies / ontologies, eg your MnM catalogue for STW Thesaurus for Economics ID (P3911). Would Wikidata:WikiProject Economics have an interest in this? They seem a bit quiet. Jheald (talk) 15:41, 4 March 2021 (UTC)

Thanks, @Jheald:, for pinging me. The STW Thesaurus for Economics (Q26903352) has a section named P Commodities. Mapping this section to Wikidata is just starting. The thesaurus has its focus on the current economy. Besides commodities, it has categories on modern transport, mapped to Wikidata already. It will be interesting to have in the end a mapping of historical and modern terms via Wikidata.

Another great source for terms on commodities in many languages is the AGROVOC (Q292649) thesaurus, which also includes a section on products. -- Jneubert (talk) 07:23, 6 March 2021 (UTC)

Cathedrals

• may be a useful group of items to look at, as we may have more detailed items for them. May give an idea of what sort of properties we should be looking for on items. Jheald (talk) 15:38, 23 February 2021 (UTC)

How do we handle Old St Paul's / New St Paul's ? Jheald (talk) 15:49, 23 February 2021 (UTC)

Item for new cathedral St Paul's Cathedral (Q173882), linked to item for old cathedral Old St Paul's Cathedral (Q2576524) by structure replaced by (P167) and structure replaces (P1398).DrThneed (talk) 04:07, 24 February 2021 (UTC)

@DrThneed: Thank you !! It may make an interesting test case for how consistently statements on St Paul's are in compliance with the contemporary constraint -- do we have a lot of medieval bishops attached to a church that wasn't built yet? How do we handle that, particularly for cases other than St Pauls (eg some grand houses), where there may be a succession of buildings, but only one article for all of them on Wikipedia, that eg links would need to be pulled to for infoboxes. Where's that "scratches head" emoji ? Jheald (talk) 11:31, 24 February 2021 (UTC)

@Jheald: There are lots of people attached to both St Pauls but whether they are all attached to the right one - I might take a look later. I don't think it helps that the items are Old St Pauls and St Pauls (not new St Pauls). Interesting point about infoboxes. So of course normally on WP pages if there are more than Wikidata item relating to the page, there are, say, two separate items with an overarching item that links to Wikipedia, e.g. individual members of a family, then a a family item linked to the family WP page. St Paul's is probably unusual in having pages for both new and old cathedrals, and thus no overarching item. My main experience with WD-fed infoboxes is people deleting ones I've put in, so not 100% sure of the workings with overarching wikidata items. Can an infobox be put on a page with a link to a specified Wikidata item that isn't the item for the page? I.e. could I have a page for the family that showed separate infoboxes for individual members of the family? Because it feels to me like that is one solution to the infobox problem of wanting information to show that might not belong in an overarching Wikidata item.DrThneed (talk) 22:40, 24 February 2021 (UTC)