Datasets:
Suggested Improvement for IOB2 Tagging Description in Dataset Card
Dear Maintainers,
Thank you for contributing this dataset to the Hugging Face library!
I have a suggestion regarding the “Dataset Summary” section of the Dataset Card. The current description contains a paragraph that could potentially lead to confusion:
The chunk tags and the named entity tags have the format I-TYPE which means that the word is inside a phrase of type TYPE.
Only if two phrases of the same type immediately follow each other, the first word of the second phrase will have tag B-TYPE to show that it starts a new phrase.
A word with tag O is not part of a phrase.
Note the dataset uses IOB2 tagging scheme, whereas the original dataset uses IOB1.
The sentence "Only if two phrases of the same type immediately follow each other, the first word of the second phrase will have tag B-TYPE to show that it starts a new phrase" seems to imply that the B-TYPE tag is used exclusively when two phrases of the same type are adjacent. However, according to the IOB2 tagging scheme, which the dataset uses, each new phrase should begin with a B-TYPE tag, regardless of whether it follows another phrase of the same type or not.
Clarifying this point in the documentation could help users better understand the tagging scheme used in the dataset.
Might I suggest a revised version for clarity:
The chunk tags and the named entity tags follow the IOB2 tagging scheme in this dataset.
In this scheme, each word within a phrase is tagged with `I-TYPE`, indicating that the word is inside a phrase of type `TYPE`.
Additionally, the first word of each new phrase starts with a `B-TYPE` tag to signify the beginning of the phrase.
A word tagged with `O` is not part of any phrase.
Note that this differs from the original dataset, which employs the IOB1 tagging scheme.
Best regards,
Ben