Upload data
Browse files- README.md +65 -0
- associations_weat.parquet +3 -0
- associations_wefat.parquet +3 -0
- words.parquet +3 -0
README.md
ADDED
@@ -0,0 +1,65 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
configs:
|
5 |
+
- config_name: words
|
6 |
+
data_files:
|
7 |
+
- split: words
|
8 |
+
path: words.parquet
|
9 |
+
- config_name: associations
|
10 |
+
data_files:
|
11 |
+
- split: associations_weat
|
12 |
+
path: associations_weat.parquet
|
13 |
+
- config_name: associations_wefat
|
14 |
+
data_files:
|
15 |
+
- split: associations_wefat
|
16 |
+
path: associations_wefat.parquet
|
17 |
+
---
|
18 |
+
|
19 |
+
# Usage
|
20 |
+
|
21 |
+
When downloading, specify which files you want to download and set the split to `train` (required by `datasets`).
|
22 |
+
|
23 |
+
```python
|
24 |
+
from datasets import load_dataset
|
25 |
+
|
26 |
+
words = load_dataset("fairnlp/weat", data_files=["words.parquet"], split="train")
|
27 |
+
associations = load_dataset("fairnlp/weat", data_files=["associations_weat.parquet"], split="train")
|
28 |
+
```
|
29 |
+
|
30 |
+
# Dataset Card for Word Embedding Association Test (WEAT)
|
31 |
+
|
32 |
+
This dataset contains the source words of the original Word Embedding Association Test (WEAT) as
|
33 |
+
described [by Caliskan et. al. (2016)](https://arxiv.org/abs/1608.07187).
|
34 |
+
|
35 |
+
## Dataset Details
|
36 |
+
|
37 |
+
The dataset contains word lists and attribute lists used to compute several WEAT scores for different embedding
|
38 |
+
associations. For details on the methodology, please refer to the original paper. This dataset is contributed to Hugging
|
39 |
+
Face as part of the WEAT implementation in the [FairNLP `fairscore` library](https://github.com/FairNLP/fairscore/).
|
40 |
+
|
41 |
+
### Dataset Sources
|
42 |
+
|
43 |
+
- **Paper [optional]:** lcs.bath.ac.uk/~jjb/ftp/CaliskanSemantics-Arxiv.pdf
|
44 |
+
|
45 |
+
**BibTeX:**
|
46 |
+
|
47 |
+
```bibtex
|
48 |
+
@article{DBLP:journals/corr/IslamBN16,
|
49 |
+
author = {Aylin Caliskan Islam and
|
50 |
+
Joanna J. Bryson and
|
51 |
+
Arvind Narayanan},
|
52 |
+
title = {Semantics derived automatically from language corpora necessarily
|
53 |
+
contain human biases},
|
54 |
+
journal = {CoRR},
|
55 |
+
volume = {abs/1608.07187},
|
56 |
+
year = {2016},
|
57 |
+
url = {http://arxiv.org/abs/1608.07187},
|
58 |
+
eprinttype = {arXiv},
|
59 |
+
eprint = {1608.07187},
|
60 |
+
timestamp = {Sat, 23 Jan 2021 01:20:12 +0100},
|
61 |
+
biburl = {https://dblp.org/rec/journals/corr/IslamBN16.bib},
|
62 |
+
bibsource = {dblp computer science bibliography, https://dblp.org}
|
63 |
+
}
|
64 |
+
```
|
65 |
+
|
associations_weat.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:16c87434655233152a9c1ee48756698274c9b64cbb60a0715bfe2d814f2532ce
|
3 |
+
size 4420
|
associations_wefat.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:89268542dbf74d3ff7a5deb6a8620740b5ccc36e8917203e60fe722af4a2c8de
|
3 |
+
size 3475
|
words.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0aa0123ad516ba495901534ed280f7f62e9a5d6380040f27f13059cf9b7a692b
|
3 |
+
size 5380
|