Datasets:
Tasks:
Text Classification
Modalities:
Text
Sub-tasks:
multi-input-text-classification
Languages:
Catalan
Size:
10K - 100K
License:
muxitox
commited on
Commit
·
31e8cfa
1
Parent(s):
e3dd34f
Reupload files
Browse files- README.md +155 -0
- dev.jsonl +0 -0
- parafraseja.py +100 -0
- test.jsonl +0 -0
- train.jsonl +0 -0
README.md
ADDED
@@ -0,0 +1,155 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
annotations_creators:
|
3 |
+
- CLiC-UB
|
4 |
+
language_creators:
|
5 |
+
- found
|
6 |
+
language:
|
7 |
+
- ca
|
8 |
+
license:
|
9 |
+
- cc-by-nc-nd-4.0
|
10 |
+
multilinguality:
|
11 |
+
- monolingual
|
12 |
+
pretty_name: Parafraseja
|
13 |
+
size_categories:
|
14 |
+
- ?
|
15 |
+
task_categories:
|
16 |
+
- text-classification
|
17 |
+
task_ids:
|
18 |
+
- multi-input-text-classification
|
19 |
+
---
|
20 |
+
|
21 |
+
# Dataset Card for Parafraseja
|
22 |
+
|
23 |
+
## Table of Contents
|
24 |
+
- [Table of Contents](#table-of-contents)
|
25 |
+
- [Dataset Description](#dataset-description)
|
26 |
+
- [Dataset Summary](#dataset-summary)
|
27 |
+
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
28 |
+
- [Languages](#languages)
|
29 |
+
- [Dataset Structure](#dataset-structure)
|
30 |
+
- [Data Instances](#data-instances)
|
31 |
+
- [Data Fields](#data-fields)
|
32 |
+
- [Data Splits](#data-splits)
|
33 |
+
- [Dataset Creation](#dataset-creation)
|
34 |
+
- [Curation Rationale](#curation-rationale)
|
35 |
+
- [Annotations](#annotations)
|
36 |
+
- [Personal and Sensitive Information](#personal-and-sensitive-information)
|
37 |
+
- [Considerations for Using the Data](#considerations-for-using-the-data)
|
38 |
+
- [Social Impact of Dataset](#social-impact-of-dataset)
|
39 |
+
- [Discussion of Biases](#discussion-of-biases)
|
40 |
+
- [Other Known Limitations](#other-known-limitations)
|
41 |
+
- [Additional Information](#additional-information)
|
42 |
+
- [Dataset Curators](#dataset-curators)
|
43 |
+
- [Licensing Information](#licensing-information)
|
44 |
+
- [Citation Information](#citation-information)
|
45 |
+
- [Contributions](#contributions)
|
46 |
+
|
47 |
+
## Dataset Description
|
48 |
+
|
49 |
+
- **Point of Contact:** [blanca.calvo@bsc.es](blanca.calvo@bsc.es)
|
50 |
+
|
51 |
+
### Dataset Summary
|
52 |
+
|
53 |
+
Parafraseja is a dataset of 21,984 pairs of sentences with a label that indicates if they are paraphrases or not. The original sentences were collected from [TE-ca](https://huggingface.co/datasets/projecte-aina/teca) and [STS-ca](https://huggingface.co/datasets/projecte-aina/sts-ca). For each sentence, an annotator wrote a sentence that was a paraphrase and another that was not. The guidelines of this annotation are available.
|
54 |
+
|
55 |
+
### Supported Tasks and Leaderboards
|
56 |
+
|
57 |
+
This dataset is mainly intended to train models for paraphrase detection.
|
58 |
+
|
59 |
+
### Languages
|
60 |
+
|
61 |
+
The dataset is in Catalan (`ca-CA`).
|
62 |
+
|
63 |
+
## Dataset Structure
|
64 |
+
|
65 |
+
The dataset consists of pairs of sentences labelled with "Parafrasis" or "No Parafrasis" in a jsonl format.
|
66 |
+
|
67 |
+
### Data Instances
|
68 |
+
|
69 |
+
<pre>
|
70 |
+
{
|
71 |
+
"id": "te1_14977_1",
|
72 |
+
"source": "teca",
|
73 |
+
"original": "La 2a part consta de 23 cap\u00edtols, cadascun dels quals descriu un ocell diferent.",
|
74 |
+
"new": "La segona part consisteix en vint-i-tres cap\u00edtols, cada un dels quals descriu un ocell diferent.",
|
75 |
+
"label": "Parafrasis"
|
76 |
+
}
|
77 |
+
</pre>
|
78 |
+
|
79 |
+
### Data Fields
|
80 |
+
- original: original sentence
|
81 |
+
- new: new sentence, which could be a paraphrase or a non-paraphrase
|
82 |
+
- label: relation between original and new
|
83 |
+
|
84 |
+
### Data Splits
|
85 |
+
|
86 |
+
* dev.json: 2,000 examples
|
87 |
+
* test.json: 4,000 examples
|
88 |
+
* train.json: 15,984 examples
|
89 |
+
|
90 |
+
## Dataset Creation
|
91 |
+
|
92 |
+
### Curation Rationale
|
93 |
+
|
94 |
+
We created this corpus to contribute to the development of language models in Catalan, a low-resource language.
|
95 |
+
|
96 |
+
### Source Data
|
97 |
+
|
98 |
+
The original sentences of this dataset came from the [STS-ca](https://huggingface.co/datasets/projecte-aina/sts-ca) and the [TE-ca](https://huggingface.co/datasets/projecte-aina/teca).
|
99 |
+
|
100 |
+
#### Initial Data Collection and Normalization
|
101 |
+
|
102 |
+
11,543 of the original sentences came from TE-ca, and 10,441 came from STS-ca.
|
103 |
+
|
104 |
+
#### Who are the source language producers?
|
105 |
+
|
106 |
+
TE-ca and STS-ca come from the [Catalan Textual Corpus](https://zenodo.org/record/4519349#.Y1Zs__uxXJF), which consists of several corpora gathered from web crawling and public corpora, and [Vilaweb](https://www.vilaweb.cat), a Catalan newswire.
|
107 |
+
|
108 |
+
### Annotations
|
109 |
+
|
110 |
+
The dataset is annotated with the label "Parafrasis" or "No Parafrasis" for each pair of sentences.
|
111 |
+
|
112 |
+
#### Annotation process
|
113 |
+
|
114 |
+
The annotation process was done by a single annotator and reviewed by another.
|
115 |
+
|
116 |
+
#### Who are the annotators?
|
117 |
+
|
118 |
+
The annotators were Catalan native speakers, with a background on linguistics.
|
119 |
+
|
120 |
+
### Personal and Sensitive Information
|
121 |
+
|
122 |
+
No personal or sensitive information included.
|
123 |
+
|
124 |
+
## Considerations for Using the Data
|
125 |
+
|
126 |
+
### Social Impact of Dataset
|
127 |
+
|
128 |
+
We hope this corpus contributes to the development of language models in Catalan, a low-resource language.
|
129 |
+
|
130 |
+
### Discussion of Biases
|
131 |
+
|
132 |
+
We are aware that this data might contain biases. We have not applied any steps to reduce their impact.
|
133 |
+
|
134 |
+
### Other Known Limitations
|
135 |
+
|
136 |
+
[N/A]
|
137 |
+
|
138 |
+
## Additional Information
|
139 |
+
|
140 |
+
### Dataset Curators
|
141 |
+
|
142 |
+
Text Mining Unit (TeMU) at the Barcelona Supercomputing Center (bsc-temu@bsc.es)
|
143 |
+
|
144 |
+
|
145 |
+
This work was funded by the [Departament de la Vicepresidència i de Polítiques Digitals i Territori de la Generalitat de Catalunya](https://politiquesdigitals.gencat.cat/ca/inici/index.html#googtrans(ca|en) within the framework of [Projecte AINA](https://politiquesdigitals.gencat.cat/ca/economia/catalonia-ai/aina).
|
146 |
+
|
147 |
+
### Licensing Information
|
148 |
+
|
149 |
+
[Creative Commons Attribution Non-commercial No-Derivatives 4.0 International](https://creativecommons.org/licenses/by-nc-nd/4.0/).
|
150 |
+
|
151 |
+
|
152 |
+
|
153 |
+
### Contributions
|
154 |
+
|
155 |
+
[N/A]
|
dev.jsonl
ADDED
The diff for this file is too large to render.
See raw diff
|
|
parafraseja.py
ADDED
@@ -0,0 +1,100 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
# Loading script for the ReviewsFinder dataset.
|
2 |
+
|
3 |
+
|
4 |
+
import json
|
5 |
+
import csv
|
6 |
+
|
7 |
+
import datasets
|
8 |
+
|
9 |
+
|
10 |
+
logger = datasets.logging.get_logger(__name__)
|
11 |
+
|
12 |
+
|
13 |
+
_CITATION = """ """
|
14 |
+
|
15 |
+
|
16 |
+
_DESCRIPTION = """ Parafraseja is a dataset of 16,584 pairs of sentences with a label that indicates if they are paraphrases or not. The original sentences were collected from TE-ca and STS-ca. For each sentence, an annotator wrote a sentence that was a paraphrase and another that was not. The guidelines of this annotation are available. """
|
17 |
+
|
18 |
+
|
19 |
+
_HOMEPAGE = """ https://huggingface.co/datasets/projecte-aina/Parafraseja/ """
|
20 |
+
|
21 |
+
|
22 |
+
|
23 |
+
_URL = "https://huggingface.co/datasets/projecte-aina/Parafraseja/resolve/main/"
|
24 |
+
_TRAINING_FILE = "train.jsonl"
|
25 |
+
_DEV_FILE = "dev.jsonl"
|
26 |
+
_TEST_FILE = "test.jsonl"
|
27 |
+
|
28 |
+
|
29 |
+
class ParafrasejaConfig(datasets.BuilderConfig):
|
30 |
+
""" Builder config for the Parafraseja dataset """
|
31 |
+
|
32 |
+
def __init__(self, **kwargs):
|
33 |
+
"""BuilderConfig for parafrasis.
|
34 |
+
Args:
|
35 |
+
**kwargs: keyword arguments forwarded to super.
|
36 |
+
"""
|
37 |
+
super(ParafrasejaConfig, self).__init__(**kwargs)
|
38 |
+
|
39 |
+
|
40 |
+
class Parafraseja(datasets.GeneratorBasedBuilder):
|
41 |
+
""" Parafrasis Dataset """
|
42 |
+
|
43 |
+
|
44 |
+
BUILDER_CONFIGS = [
|
45 |
+
ParafrasejaConfig(
|
46 |
+
name="Parafraseja",
|
47 |
+
version=datasets.Version("1.0.0"),
|
48 |
+
description="Parafraseja dataset",
|
49 |
+
),
|
50 |
+
]
|
51 |
+
|
52 |
+
|
53 |
+
def _info(self):
|
54 |
+
return datasets.DatasetInfo(
|
55 |
+
description=_DESCRIPTION,
|
56 |
+
features=datasets.Features(
|
57 |
+
{
|
58 |
+
"sentence1": datasets.Value("string"),
|
59 |
+
"sentence2": datasets.Value("string"),
|
60 |
+
"label": datasets.features.ClassLabel
|
61 |
+
(names=
|
62 |
+
[
|
63 |
+
"No Parafrasis",
|
64 |
+
"Parafrasis",
|
65 |
+
]
|
66 |
+
),
|
67 |
+
}
|
68 |
+
),
|
69 |
+
homepage=_HOMEPAGE,
|
70 |
+
citation=_CITATION,
|
71 |
+
)
|
72 |
+
|
73 |
+
|
74 |
+
def _split_generators(self, dl_manager):
|
75 |
+
"""Returns SplitGenerators."""
|
76 |
+
urls_to_download = {
|
77 |
+
"train": f"{_URL}{_TRAINING_FILE}",
|
78 |
+
"dev": f"{_URL}{_DEV_FILE}",
|
79 |
+
"test": f"{_URL}{_TEST_FILE}",
|
80 |
+
}
|
81 |
+
downloaded_files = dl_manager.download_and_extract(urls_to_download)
|
82 |
+
|
83 |
+
return [
|
84 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloaded_files["train"]}),
|
85 |
+
datasets.SplitGenerator(name=datasets.Split.VALIDATION, gen_kwargs={"filepath": downloaded_files["dev"]}),
|
86 |
+
datasets.SplitGenerator(name=datasets.Split.TEST, gen_kwargs={"filepath": downloaded_files["test"]}),
|
87 |
+
]
|
88 |
+
|
89 |
+
|
90 |
+
def _generate_examples(self, filepath):
|
91 |
+
"""This function returns the examples in the raw (text) form."""
|
92 |
+
logger.info("generating examples from = %s", filepath)
|
93 |
+
with open(filepath, encoding="utf-8") as f:
|
94 |
+
data = [json.loads(line) for line in f]
|
95 |
+
for id_, article in enumerate(data):
|
96 |
+
yield id_, {
|
97 |
+
"sentence1": article['original'],
|
98 |
+
"sentence2": article['new'],
|
99 |
+
"label": article['label'],
|
100 |
+
}
|
test.jsonl
ADDED
The diff for this file is too large to render.
See raw diff
|
|
train.jsonl
ADDED
The diff for this file is too large to render.
See raw diff
|
|