File size: 4,932 Bytes
0142de9 8cdbc47 3b85975 4ddde2b 0142de9 fd41e53 61afd7c 0142de9 81c5259 0142de9 8229163 0142de9 63ecc61 4d4f1f0 63ecc61 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 | ---
license: cc-by-nc-4.0
language:
- de
tags:
- gender-fair language
pretty_name: Lou
size_categories:
- n<1K
configs:
- config_name: germeval-toxic
data_files:
- split: test
path: "germeval-toxic.jsonl"
- config_name: germeval-factclaiming
data_files:
- split: test
path: "germeval-factclaiming.jsonl"
- config_name: germeval-engaging
data_files:
- split: test
path: "germeval-engaging.jsonl"
- config_name: x-stance-de
data_files:
- split: test
path: "x-stance-de.jsonl"
---
## The Lou Dataset
*Figure 1, example entry of the `Lou` dataset with the `original` instance of the engaging detection task from the GermEval-2021 dataset and its six reformulations.*
The `Lou` dataset provides gender-fair reformulations for instances from seven German classification tasks.
It is intended for non-commercial use and research is licensed under the CC BY-NC 4.0 licence.
The copyright of the original text remains at the original datasets.
### Tasks and Data
We include seven tasks (*sentiment analysis* and *stance-*, *fact-claiming-*, *engaging-*, *hate-speech-*, and *toxicity-detection*) from the [X-Stance](https://huggingface.co/datasets/ZurichNLP/x_stance), [GermEval-2021](https://germeval2021toxic.github.io/SharedTask/), and [DeTox](https://github.com/hdaSprachtechnologie/detox) datasets.
For `Lou`, you need access to the full version of DeTox.
Therefore, we excluded it from this public repository.
However, we are happy to share this part of `Lou` with you when you provide us the approval, which you can request [here](https://github.com/hdaSprachtechnologie/detox).
### Reformulation Strategies
With `Lou`, we provide gender-inclusive and gender-neutral reformulations for masculine formulations, like *Politiker* in *Figure 1*.
We use six strategies: `Doppelnennung`, `GenderStern`, `GenderDoppelpunkt`, `GenderGap`, `Neutral`, and `De-e`:
* **Binary Gender Inclusion** (`Doppelnennung`) explicitly mentions the feminine and masculine but ignores others like agender.
For example, *Politiker* (politician.MASC.PL) is transformed into *Politikerinnen und Politiker* (*politician.FEM.PL and politician.MASC.PL*).
* **All Gender Inclusion** explicitly addresses every gender, including agender, non-binary, or demi-gender, using a gender gap character pronounced with a small pause.
We consider three commonly used strategies with different gender characters: `GenderStern` (\*), `GenderDoppelpunkt` (:), and `GenderGap` (_).
For example, *Politiker* (politician.MASC.PL) is turned into *Politiker*innen*, *Politiker:innen*, or *Politiker_innen* (*politician.FEM.MASC.NEUT.PL*).
* **Gender Neutralization** avoids naming a particular gender. For this strategy (`Neutral`), we use neutral terms like *ärztliche Fachperson* (*medical professional*).
* **Neosystem** (`De-e`) is a well-specified system that emerged from a significant [community-driven effort](https://geschlechtsneutral.net).
This strategy uses a fourth gender, including new pronouns, articles, and suffixes to avoid naming a particular gender.
For example, *Politiker* (*politician.MASC.PL*) is changed to *Politikerne* (*politician.FEM.MASC.NEUT.PL*).
### Reference
```
@inproceedings{waldis-etal-2024-lou,
title = "The {L}ou Dataset - Exploring the Impact of Gender-Fair Language in {G}erman Text Classification",
author = "Waldis, Andreas and
Birrer, Joel and
Lauscher, Anne and
Gurevych, Iryna",
editor = "Al-Onaizan, Yaser and
Bansal, Mohit and
Chen, Yun-Nung",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
month = nov,
year = "2024",
address = "Miami, Florida, USA",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2024.emnlp-main.592",
pages = "10604--10624",
abstract = "Gender-fair language, an evolving linguistic variation in German, fosters inclusion by addressing all genders or using neutral forms. However, there is a notable lack of resources to assess the impact of this language shift on language models (LMs) might not been trained on examples of this variation. Addressing this gap, we present Lou, the first dataset providing high-quality reformulations for German text classification covering seven tasks, like stance detection and toxicity classification. We evaluate 16 mono- and multi-lingual LMs and find substantial label flips, reduced prediction certainty, and significantly altered attention patterns. However, existing evaluations remain valid, as LM rankings are consistent across original and reformulated instances. Our study provides initial insights into the impact of gender-fair language on classification for German. However, these findings are likely transferable to other languages, as we found consistent patterns in multi-lingual and English LMs.",
}
```
|