Datasets:
Add information for the dataset
Browse files
README.md
CHANGED
@@ -30,6 +30,10 @@ This dataset contains:
|
|
30 |
|
31 |
* The [annotation guidelines](#annotation-guidelines) that have been used for building and curating the dataset, if they've been defined in Argilla.
|
32 |
|
|
|
|
|
|
|
|
|
33 |
### Load with Argilla
|
34 |
|
35 |
To load with Argilla, you'll just need to install Argilla as `pip install argilla --upgrade` and then use the following code:
|
@@ -265,6 +269,119 @@ The dataset contains a single split, which is `train`.
|
|
265 |
|
266 |
## Dataset Creation
|
267 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
268 |
### Curation Rationale
|
269 |
|
270 |
[More Information Needed]
|
|
|
30 |
|
31 |
* The [annotation guidelines](#annotation-guidelines) that have been used for building and curating the dataset, if they've been defined in Argilla.
|
32 |
|
33 |
+
It contains the raw version of [go_emotions](https://huggingface.co/datasets/go_emotions) as a `FeedbackDataset`. Each of the original questions are defined a single
|
34 |
+
`FeedbackRecord` and contain the `responses` from each annotator. The final labels in the *simplified* version of the dataset have been used as `suggestions`, so that we
|
35 |
+
can use this dataset to showcase the metrics related to the agreement between annotators as well as the `responses` vs `suggestions` metrics.
|
36 |
+
|
37 |
### Load with Argilla
|
38 |
|
39 |
To load with Argilla, you'll just need to install Argilla as `pip install argilla --upgrade` and then use the following code:
|
|
|
269 |
|
270 |
## Dataset Creation
|
271 |
|
272 |
+
### Script used for the generation
|
273 |
+
|
274 |
+
```python
|
275 |
+
import argilla as rg
|
276 |
+
from datasets import load_dataset
|
277 |
+
import uuid
|
278 |
+
from datasets import concatenate_datasets
|
279 |
+
|
280 |
+
ds = load_dataset("go_emotions", "raw", split="train")
|
281 |
+
ds_prepared = load_dataset("go_emotions")
|
282 |
+
|
283 |
+
_CLASS_NAMES = [
|
284 |
+
"admiration",
|
285 |
+
"amusement",
|
286 |
+
"anger",
|
287 |
+
"annoyance",
|
288 |
+
"approval",
|
289 |
+
"caring",
|
290 |
+
"confusion",
|
291 |
+
"curiosity",
|
292 |
+
"desire",
|
293 |
+
"disappointment",
|
294 |
+
"disapproval",
|
295 |
+
"disgust",
|
296 |
+
"embarrassment",
|
297 |
+
"excitement",
|
298 |
+
"fear",
|
299 |
+
"gratitude",
|
300 |
+
"grief",
|
301 |
+
"joy",
|
302 |
+
"love",
|
303 |
+
"nervousness",
|
304 |
+
"optimism",
|
305 |
+
"pride",
|
306 |
+
"realization",
|
307 |
+
"relief",
|
308 |
+
"remorse",
|
309 |
+
"sadness",
|
310 |
+
"surprise",
|
311 |
+
"neutral",
|
312 |
+
]
|
313 |
+
label_to_id = {label: i for i, label in enumerate(_CLASS_NAMES)}
|
314 |
+
id_to_label = {i: label for i, label in enumerate(_CLASS_NAMES)}
|
315 |
+
|
316 |
+
# Concatenate the datasets and transform to pd.DataFrame
|
317 |
+
|
318 |
+
ds_prepared = concatenate_datasets([ds_prepared["train"], ds_prepared["validation"], ds_prepared["test"]])
|
319 |
+
df_prepared = ds_prepared.to_pandas()
|
320 |
+
|
321 |
+
# Obtain the final labels as a dict, to later include these as suggestions
|
322 |
+
|
323 |
+
labels_prepared = {}
|
324 |
+
for idx in df_prepared.index:
|
325 |
+
labels = [id_to_label[label_id] for label_id in df_prepared['labels'][idx]]
|
326 |
+
labels_prepared[df_prepared['id'][idx]] = labels
|
327 |
+
|
328 |
+
# Add labels to the dataset and keep only the relevant columns
|
329 |
+
|
330 |
+
def add_labels(ex):
|
331 |
+
labels = []
|
332 |
+
for label in _CLASS_NAMES:
|
333 |
+
if ex[label] == 1:
|
334 |
+
labels.append(label)
|
335 |
+
ex["labels"] = labels
|
336 |
+
|
337 |
+
return ex
|
338 |
+
|
339 |
+
ds = ds.map(add_labels)
|
340 |
+
df = ds.select_columns(["text", "labels", "rater_id", "id"]).to_pandas()
|
341 |
+
|
342 |
+
# Create a FeedbackDataset for text classification
|
343 |
+
|
344 |
+
feedback_dataset = rg.FeedbackDataset.for_text_classification(labels=_CLASS_NAMES, multi_label=True)
|
345 |
+
|
346 |
+
# Create the records with the original responses, and use as suggestions
|
347 |
+
# the final labels in the "simplified" go_emotions dataset.
|
348 |
+
|
349 |
+
records = []
|
350 |
+
for text, df_text in df.groupby("text"):
|
351 |
+
responses = []
|
352 |
+
for rater_id, df_raters in df_text.groupby("rater_id"):
|
353 |
+
responses.append(
|
354 |
+
{
|
355 |
+
"values": {"label": {"value": df_raters["labels"].iloc[0].tolist()}},
|
356 |
+
"status": "submitted",
|
357 |
+
"user_id": uuid.UUID(int=rater_id),
|
358 |
+
}
|
359 |
+
)
|
360 |
+
suggested_labels = labels_prepared.get(df_raters["id"].iloc[0], None)
|
361 |
+
if not suggested_labels:
|
362 |
+
continue
|
363 |
+
suggestion = [
|
364 |
+
{
|
365 |
+
"question_name": "label",
|
366 |
+
"value": suggested_labels,
|
367 |
+
"type": "human",
|
368 |
+
}
|
369 |
+
]
|
370 |
+
records.append(
|
371 |
+
rg.FeedbackRecord(
|
372 |
+
fields={"text": df_raters["text"].iloc[0]},
|
373 |
+
responses=responses,
|
374 |
+
suggestions=suggestion
|
375 |
+
)
|
376 |
+
)
|
377 |
+
|
378 |
+
|
379 |
+
feedback_dataset.add_records(records)
|
380 |
+
|
381 |
+
# Push to the hub
|
382 |
+
feedback_dataset.push_to_huggingface("plaguss/go_emotions_raw")
|
383 |
+
```
|
384 |
+
|
385 |
### Curation Rationale
|
386 |
|
387 |
[More Information Needed]
|