`cr:FileObject` not recognized by the validation script

[ihsaan-ullah]

I have a metadata.json file with minimal croissant metadata for a dataset. The dataset has multiple zip files for which I am using cr:FileObject. When I validate the metadata.json file using mlcroissant python package, I get error for FileObject.

Here is how I validate my metadata.json

mlcroissant validate --jsonld metadata.json

Here is the error I get for FileObjec:

Found the following 9 error(s) during the validation:
  -  "ICLR2024_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "ICLR2025_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "ICML2024_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "ICML2025_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "NeurIPS2021_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "NeurIPS2022_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "NeurIPS2023_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  "NeurIPS2024_latest.zip" should have an attribute "@type": "https://schema.org/FileObject" or "@type": "https://schema.org/FileSet". Got https://mlcommons.org/croissant/1.0/FileObject instead.
  -  [Metadata(Academic Papers Dataset)] The name "Academic Papers Dataset" contains forbidden characters.

The last error about name is also interesting, what forbidden characters are there in the name? Maybe Dataset

Here is my complete metadata.json

[benjelloun]

The error message is strange, but I do see a mistake in the definition of the namespace:

"cr": "https://mlcommons.org/croissant/1.0/" 

Should be

"cr": "https://mlcommons.org/croissant/" 

As the namespace is not versioned.

Also, you probably want to copy the standard context from the bottom of the spec into your dataset.