Closed ColmMassey closed 4 years ago
As of the morning of 12/12 we can load dotcoop (12/04/2019), or co-opsuk(q3/2019) or both. Looking closely at the Edinburgh Bicycle co-op (in Edinburgh) I'm seeing the following..
In the Co-ops UK graph, there is one entry for the co-op and one for the outlet. They have the same postcode, but the rest of the address has differences. The outlet also has a description. Neither has a website. Both have sameas links to the dot-coop uri.
In the DotCoop graph, there is only one for the co-op. It has the same postcode as the Co-ops UK data, but the rest of the address has difference to both the Co-ops UK entries. It has a website. It has geo data and a URI for ordanance survey. It has sameas links to both coopsuk entries.
This looks exactly as I would expect. :-)
If I switch to just using the DotCo-op data it still presents to icons for Edinburgh Bicycle co-op in Edinburgh. The data is identical. Any ideas why this might be the case?
Toggling back to the co-ops-uk data set, I can't find the Edinburgh Bicycle co-op at all?
To merge the data we first take the duplicated ones between each table and merge them. Then we take the entries that do not contain a sameas predicate from one table. Then we take the entries that do not contain a sameas predicate from the other table.
We are hard-coding the merge currently and have not made it generic (i.e. you can merge only coops uk and dotcoop datasets). We also have hard-coded the way the merging works, we are currently taking the following fields from the tables:
coopsuk
dotcoop
merged results
Something like this https://sketch.cloud/s/75lep/a/94nKMM/play