Datasets:
add up to social impact and biases
Browse files
README.md
CHANGED
@@ -290,4 +290,18 @@ We kept preprocessing minimal so that the process was as language agnostic as po
|
|
290 |
|
291 |
## Considerations for Using the Data
|
292 |
|
293 |
-
### Social Impact of Dataset
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
290 |
|
291 |
## Considerations for Using the Data
|
292 |
|
293 |
+
### Social Impact of Dataset
|
294 |
+
|
295 |
+
This dataset covers a number of low-resourced languages. This makes it a potentially useful resource, but due to the limited amount of data and domains, care must be taken not to overclaim performance or coverage.
|
296 |
+
|
297 |
+
### Discussion of Biases
|
298 |
+
|
299 |
+
Our work aims to broaden natural language processing coverage by allowing practitioners to identify relevant data in more languages. However, we note that language identification is inherently a normative activity that risks excluding minority dialects, scripts, or entire microlanguages from a macrolanguage. Choosing which languages to cover may reinforce power imbalances, as only some groups gain access to language processing technologies.
|
300 |
+
|
301 |
+
In addition, errors in language identification can have a significant impact on downstream performance, particularly (as is often the case) when a system is used as a `black box'. The performance of our classifier is not equal across languages which could lead to worse downstream performance for particular groups. We mitigate this by providing metrics by class.
|
302 |
+
|
303 |
+
### Licensing Information
|
304 |
+
|
305 |
+
### Citation Information
|
306 |
+
|
307 |
+
### Contributions
|