Closed jyucsiro closed 4 years ago
Yep, looks like a bug! Good catch.
Edit: Or maybe not? This is the relevant section of code, it seems right: https://github.com/CSIRO-enviro-informatics/asgs-dataset/blob/61cd90d5487a80bc902d2cbbb38645093542b1da/asgs_dataset/model/asgs_feature.py#L1555
I've checked the ttl output for both SUA and RA on asgsld.net and both looks to be producing rdf types with correct prefix.
it's strange though that it's appearing in the cache. @benjaminleighton any ideas?
I'm not sure but a place to check would be in the dataset downloaded from s3 as part of the cache build, the exact reference will be part of download-data.sh on the cache machine itself but @ashleysommer might have a good idea which file to look in.
@jyucsiro
Yep, that looks like the bug that causes it. Its actually part of the SA2 feature RDF mapping. When a SA2 is part of a SUA, then it puts /remotenessarea/1000 a SUA
and this.sa2 sfWithin /remotenessarea/1000
Instead should be /significanturbanarea/1000 a SUA
and this.sa2 sfWithin /significanturbanarea/1000
So actually what needs to be reharvested here is the set of SA2s. Thats why I didn't find the problematic triples when looking in the SUA set and the RA set.
I've fixed the bug and reharvesting set of SA2s now.
To fix the current cache, we'll have to remove the any instances of triples which say /remotenessarea/x rdf:type asgs#SUA
and any that say this.sa2 sfWithin /remotenessarea/x
Then re-ingest the new set of SA2s with the correct sfWithin /significanturbanarea/x
triples
Hitting into some confusing URIs. RA URI prefixes are used in SUAs
Running a query like this on the triple store cache:
Excerpt of the result of the query:
For some reason, SUAs have the same prefix as remoteness areas... is this a bug?