I was under the impression that each time the claim is updated, the new ID is generated, so I can just match the IDs to update the claims. However, I have discovered the following: for item Q24517, the old claim in my dump is:
"P279":[{"id":"Q24517$7A855A08-5DD0-41A6-9A36-E3AC3DE24B11","mainsnak":{"snaktype":"value","property":"P279","datatype":"wikibase-item","datavalue":{"value":{"entity-type":"item","numeric-id":2095},"type":"wikibase-entityid"}}However, in current dump at https://www.wikidata.org/wiki/Special:EntityData/Q24517.json it is:
"P279":[{"id":"Q24517$7A855A08-5DD0-41A6-9A36-E3AC3DE24B11","mainsnak":{"snaktype":"value","property":"P279","datatype":"wikibase-item","datavalue":{"value":{"entity-type":"item","numeric-id":2207288},"type":"wikibase-entityid"}}As we can see, same claim ID but it refers to a different node. That makes it hard to recognize when the claim must be updated. So I'd like to figure out:
- Is it intentional or a bug?
- If it's intentional, can it be changed to generate new IDs on change?
- If not, what would be the best way to recognize when clam changes?
The items can have many claims, so knowing which ones changed and updating only those would greatly speed up the query service function.
Alternatively, if there's some other format that is better suitable to loading updates, we may want to use that instead of JSON data.