Upload greek-legal-code dataset #2966
Conversation
christospi
force-pushed
the
add-greek-legal-code-dataset
branch
from
f5cfdd5
to
775e71b
Compare
| # License for the dataset if available | ||
| license=_LICENSE, | ||
| # Citation for the dataset | ||
| citation=_CITATION, |
There was a problem hiding this comment.
we can add a text classification template:
| citation=_CITATION, | |
| citation=_CITATION, | |
| task_templates=[TextClassification(text_column="text", label_column="label")], |
| features = datasets.Features( | ||
| { | ||
| "text": datasets.Value("string"), | ||
| "label": datasets.Value("string"), |
There was a problem hiding this comment.
it would be nice to have the list of the possible labels here (even though they can be numerous in the "subject" case):
| "label": datasets.Value("string"), | |
| "label": datasets.ClassLabel(names=[...]), |
datasets/greek_legal_code/README.md
Outdated
| `text`: (**str**) The full content of each document, which is represented by its `header` and `articles` (i.e., the `main_body`).\ | ||
| `volume`: (**str**) The volume-level class it belongs to.\ | ||
| `chapter`: (**str**) The chapter-level class it belongs to.\ | ||
| `subject`: (**str**) The subject-level class it belongs to. |
There was a problem hiding this comment.
maybe we can list the name of the classes here ? Or at least for the "volume" case ?
There was a problem hiding this comment.
Thanks for adding this dataset @christospi.
I found a potential issue (see description below).
| data = json.loads(row) | ||
| yield id_, { | ||
| "text": data["text"], | ||
| "label": data[self.config.label_type], |
There was a problem hiding this comment.
I think there is an issue in the "label" field: the dataset will be identically repeated 3 times, once for each of the configuration values.
There was a problem hiding this comment.
Hi @albertvillanova, thanks for reviewing.
Why is this an issue?
The goal is that the users will be able to use 3 different configurations of the very same dataset based on the examined label set they prefer.
e.g.,
from datasets import load_dataset
# Load the dataset including the volume labels
dataset = load_dataset('greek_legal_code', label_type='volume')
# Load the dataset including the chapter labels
dataset = load_dataset('greek_legal_code', label_type='chapter')
# Load the dataset including the chapter labels
dataset = load_dataset('greek_legal_code', label_type='subject')There was a problem hiding this comment.
I think it's fine this way. The texts are the same and just the labels are changing depending on the granularity that you need for classification.
christospi
force-pushed
the
add-greek-legal-code-dataset
branch
from
ab2d82a
to
2ecbeac
Compare
|
@albertvillanova @lhoestq thank you very much for reviewing! 🤗 I 've pushed some updates/changes as requested. |
No description provided.