Fix discrepancies with the specs #742
Conversation
|
MLCommons CLA bot All contributors have signed the MLCommons CLA ✍️ ✅ |
ccl-core
changed the title
| json_copy = json_ld.copy() | ||
| for _, record_set in enumerate(json_copy.get("recordSet", [])): | ||
| if record_set["@id"] != record_set["name"]: | ||
| record_set["name"] = record_set["@id"] |
There was a problem hiding this comment.
nit: this is not a requirement of the specs.
There was a problem hiding this comment.
You are right, it is not a requirement... but it seems to be the recommended way though, given that it is so for all given examples :)
| {"name": "person7", "age": 7} | ||
| {"name": "person8", "age": 8} | ||
| {"name": "person9", "age": 9} | ||
| {"persons/name": "person0", "persons/age": 0} |
There was a problem hiding this comment.
I think it was feature to not have the RecordSet ID here - otherwise, it's repeated:
- The user is asking for the
personsRecordSet - => All fields will start with
persons
Maybe, it's a consequence of my other comment where you set name == @id for all datasets. It's the case for Hugging Face datasets, but it may also not be the case per the specs.
So it could be good to keep at least one dataset with name != @id for testing purposes. What do you think?
There was a problem hiding this comment.
Sure! I updated all datasets because of the migration script, but good point to keep one different.
There was a problem hiding this comment.
Updated comment in the migration script and restore one dataset with names != ids.
get_columnmethod ofSource, we return the node'suuidif no extract method is specified;data, we look atfield.idand notfield.nameto get the expected keys;