adding motivations
Browse files
README.md
CHANGED
|
@@ -40,6 +40,8 @@ configs:
|
|
| 40 |
|
| 41 |
**This dataset contains over 1 million excepts from U.S. Congressional Records that contain a dogwhistle term, created via keyword search on terms in the [Allen AI Dogwhistle Glossary](https://dogwhistles.allen.ai/)**. Therefore, many of the excerpts are likely using these terms in a non-coded sense. This dataset was used as input to our approach for word-sense disambiguation of dog whistles from standard speech using Large Language Models (LLMs). Please see the paper linked below for more details.
|
| 42 |
|
|
|
|
|
|
|
| 43 |
<p style="color:red;">Please note, this dataset contains content that may be upsetting or offensive to some readers.</p>
|
| 44 |
|
| 45 |
**Published at ACL 2024!**
|
|
|
|
| 40 |
|
| 41 |
**This dataset contains over 1 million excepts from U.S. Congressional Records that contain a dogwhistle term, created via keyword search on terms in the [Allen AI Dogwhistle Glossary](https://dogwhistles.allen.ai/)**. Therefore, many of the excerpts are likely using these terms in a non-coded sense. This dataset was used as input to our approach for word-sense disambiguation of dog whistles from standard speech using Large Language Models (LLMs). Please see the paper linked below for more details.
|
| 42 |
|
| 43 |
+
*Although we collect over 7 million potential dog whistle instances, due to resource constraints, we only sample 100,000 instances for the creation of the Silent Signals dataset. **Therefore, we release the Potential Dog Whistle Dataset to enable the open sourced expansion of the Silent Signals dataset.***
|
| 44 |
+
|
| 45 |
<p style="color:red;">Please note, this dataset contains content that may be upsetting or offensive to some readers.</p>
|
| 46 |
|
| 47 |
**Published at ACL 2024!**
|