User:Daniel Mietchen/ShEx for clinical trials
Jump to navigation
Jump to search
About[edit]
I am playing with Shape Expressions here and want to create a ShEx manifest for clinical trials.
Process[edit]
Here are the steps I followed:
- Explore some individual clinical trials
- Explore properties currently being used on Wikidata items on clinical trials
- as per https://w.wiki/MJc
- note - desired query would be one that counts the number of times a property is used among all clinical trials, right? to rank the most popular?
- Yes, and I am using https://w.wiki/MJv for that at the bottom
- note - desired query would be one that counts the number of times a property is used among all clinical trials, right? to rank the most popular?
- as per https://w.wiki/MJc
- Explore (read, validate, check) some of the Entity Schemas already existing on Wikidata
- as per User:HakanIST/EntitySchemaList, perusing ShEx Primer and ShEx Specification and the ShEx2 Validator as needed
- Review how Wikidata Shape Expressions Inference works
- Running Wikidata Shape Expressions Inference for a random set of clinical trial items
- https://tools.wmflabs.org/wd-shex-infer/job/56
- while waiting for the job to finish, set up
{{EntitySchema}}
based on{{P}}
in order to list schema examples conveniently above
- while waiting for the job to finish, set up
- The resulting ShEx is way too complex, so I went for a variant based on the above query for properties currently being used
- https://tools.wmflabs.org/wd-shex-infer/job/56
- Validation via ShEx2 Validator
- that gave errors once the ShEx got too long, so I had to drop some of the lesser used properties
Draft[edit]
Notes[edit]
- The result can now be compared to a more comprehensive variant of the above query for properties currently being used
- I also tested YASHE on the way and will probably look into it some more.