LCTGM Wikidata test set

[glenrobson]

Issue to investigate Wikidata as a test data source. There are 697 LCTGM terms on wikidata (query) and these are present on subject objects e.g. screenplay, orangery, love etc.

These subjects are then linked to using the Depicts relation to relate images to subjects.

[glenrobson]

Limiting the query to items with subject LCTGM headings that have a image link (query) we get the following 29 headings:

Subject	Count
portrait	10709
chair	762
hug	388
river	289
tree	225
military uniform	182
bridge	157
sport	155
poet	114
railway	113
award	91
school	65
dam	63
beach	44
fortification	29
stained glass	29
column	23
bride	18
bridegroom	18
agricultural worker	18
monastery	17
pew	17
tower	16
balustrade	14
advertising	13
cargo ship	12
Serliana	11
parking lot	11
stream	11

Query to retrieve example images

[glenrobson]

Limiting this to IIIF images in Wikidata (query) we get 16 results:

Subject	Count
sport	155
river	146
poet	102
award	91
tree	86
bridge	75
dam	53
stained glass	29
school	26
bride	18
agricultural worker	18
bridegroom	18
beach	17
pew	17
chair	11
railway	11

Query to retrieve example images

[glenrobson]

Questions:

• Should we limit to IIIF images from wikidata for test?
• Do we need to have equal amount per subject?

[glenrobson]

Here is the query to get all of the Image URLs for items that can be linked to LCTGM:

https://w.wiki/37xy

Note in the summaries above I limited it to those that have over 10 images per subject but the query above doesn't have this limit so quite a few will have just a single image linked.