Malta

Large Language models (LLMs), while powerful, exhibit harmful social biases. Debiasing is often challenging due to computational costs, data constraints, and potential degradation of multi-task language capabilities. This work introduces a novel approach utilizing ChatGPT to generate synthetic training data, aiming to enhance the debiasing of LLMs. We propose two strategies: Targeted Prompting, which provides effective debiasing for known biases but necessitates prior specification of bias in question; and General Prompting, which, while slightly less effective, offers debiasing across various categories. We leverage resource-efficient LLM debiasing using adapter tuning and compare the effectiveness of our synthetic data to existing debiasing datasets. Our results reveal that: (1) ChatGPT can efficiently produce high-quality training data for debiasing other LLMs; (2) data produced via our approach surpasses existing datasets in debiasing performance while also preserving internal knowledge of a pre-trained LLM; and (3) synthetic data exhibits generalizability across categories, effectively mitigating various biases, including intersectional ones. These findings underscore the potential of synthetic data in advancing the fairness of LLMs with minimal retraining cost.

EACL 2024

ChatGPT Based Data Augmentation for Improved Parameter-Efficient Debiasing of LLMs

synthetic data generation

social bias

large language models

### Welcome to ACL 2023, the 61st Annual Meeting of the Association for Computational Linguistics! 
 The conference will be held in Toronto, Canada, July 9-14, 2023. 
Following the succession of the recent conferences in our field, ACL 2023 will adopt a hybrid format.
While the impact of Covid has considerably diminished in terms of traveling, obtaining visas to Canada
entails a very long process. Moreover, the global economic conditions pose challenges for many individuals to travel to conferences. Recognizing these circumstances, we know many participants may not be
able to attend the conference in person. Therefore, we are committed to providing a great virtual platform
so everyone has the opportunity to interact with other participants and enjoy the conference. Based on the
current registered participants, approxiately 30% have chosen to attend the conference virtually. Whether
you join us in person or virtually, we sincerely hope everyone has a remarkable conference experience. 
This General Chair’s message is where I express my gratitude to the many individuals who have made
enormous contributions to the conference over the past year.

Read [**ACL 2023 General Chair's message**](https://docs.google.com/document/d/1WobYM7norbG4dI48s75HfJoD89qgX5a_F-6U8AteLSA/edit?usp=sharing/) in full.

##### **[Conference Handbook](https://2023.aclweb.org/downloads/acl2023-handbook.pdf)**

ACL 2023

The Association for Computational Linguistics (ACL) is the premier international scientific and professional society for people working on computational problems involving human language, a field often referred to as either computational linguistics or natural language processing.

workshop paper

Welcome to the 18th Conference of the European Chapter of the Association for Computational Linguistics (EACL). Continuing its mission of expanding and involving the science community of all European countries, EACL has selected the Malta community for the 18th EACL. Considering the importance of physical interaction among researchers, the conference will be held at the Hotel Radisson Blu, St. Julians, in Malta, from 17 to 22 of March, 2024. The conference will also feature the possibility to participate virtually. As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language. 
#### [Conference Handbook](https://drive.google.com/file/d/13Wn38q0yev6U3RTgnf4Tn5T9GuAuZ-xW/view?usp=sharing) *downloadable*

To access the EACL 2024 event page you need to register [**here**](https://acl.swoogo.com/EACL2024). 

Please register!

As the flagship European conference in the field of computational linguistics, EACL welcomes European and international researchers covering a broad spectrum of research areas that are concerned with computational approaches to natural language.

Classifying discourse relations is known as a hard task, relying on complex indices. On the other hand, discourse-annotated data is scarce, especially for languages other than English: many corpora, of limited size, exist for several languages but the domain is split between different theoretical frameworks that have a huge impact on the nature of the textual spans to be linked, and the label set used. Moreover, each annotation project implements modifications compared to the theoretical background and other projects. These discrepancies hinder the development of systems taking advantage of all the available data to tackle data sparsity and work on transfer between languages is very limited, almost nonexistent between frameworks, while it could improve our understanding of some theoretical aspects and enhance many applications. In this paper, we propose the first experiments on zero-shot learning for discourse relation classification and investigate several paths in the way source data can be combined, either based on languages, frameworks, or similarity measures. We demonstrate how difficult transfer is for the task at hand, and that the most impactful factor is label set divergence, where the notion of underlying framework possibly conceals crucial disagreements.

Zero-shot learning for multilingual discourse relation classification

This study pioneers the use of synthetically generated data for training generative models in document-level text simplification of German texts. We demonstrate the effectiveness of our approach with real-world online texts. Addressing the challenge of data scarcity in language simplification, we crawled professionally simplified German texts and synthesized a corpus using GPT-4. We finetune Large Language Models with up to 13 billion parameters on this data and evaluate their performance. This paper employs various methodologies for evaluation and demonstrates the limitations of currently used rule-based metrics. Both automatic and manual evaluations reveal that our models can significantly simplify real-world online texts, indicating the potential of synthetic data in improving text simplification.

German Text Simplification: Finetuning Large Language Models with Semi-Synthetic Data

Commonly used language defines ”hate speech” as objectionable statements that may jeopardize societal harmony by singling out a group or a person based on fundamental traits (includ- ing gender, caste, or religion). Using machine learning techniques, our research focuses on identifying hate speech in social media com- ments. Using a variety of machine learning methods, we created machine learning models to detect hate speech. An approximate F1 of 0.6 was attained by the created models.

KEC AI DSNLP@LT-EDI-2024:Caste/Immigration Hate Speech Detection using Machine Learning Techniques

Dehumanization is a mental process that enables the exclusion and ill treatment of a group of people. In this paper, we present two data sets of dehumanizing text, a large, automatically collected corpus and a smaller, manually annotated data set. Both data sets include a combination of political discourse and dialogue from movie subtitles. Our methods give us a broad and varied amount of dehumanization data to work with, enabling further exploratory analysis as well as automatic classification of dehumanization patterns. Both data sets will be publicly released.

A Dataset for the Detection of Dehumanizing Language

In the field of spoken language understanding, systems like Whisper and Multilingual Massive Speech (MMS) have shown state-of-the-art performances. This study is dedicated to a comprehensive exploration of the Whisper and MMS systems, with a focus on assessing biases in automatic speech recognition (ASR) inherent to casual conversation speech specific to the Portuguese language. Our investigation encompasses various categories, including gender, age, skin tone color, and geo-location. Alongside traditional ASR evaluation metrics such as Word Error Rate (WER), we have incorporated p-value statistical significance for gender bias analysis. Furthermore, we extensively examine the impact of data distribution and empirically show that oversampling techniques alleviate such stereotypical biases. This research represents a pioneering effort in quantifying biases in the Portuguese language context through the application of MMS and Whisper, contributing to a better understanding of ASR systems' performance in multilingual settings.

The balancing Act: Unmasking and Alleviating ASR Biases in Portuguese

In this paper, we report on a new corpus of simplified German. It is recently requested from public agencies in Germany to provide information in easy language on their outlets (e.g. websites) so as to facilitate participation in society for people with low-literacy levels related to learning difficulties or low language proficiency (e.g. L2 speakers). While various rule sets and guidelines for Easy German (a specific variant of simplified German) have emerged over time, it is unclear (a) to what extent authors and other content creators, including generative AI tools consistently apply them, and (b) how adequate texts in authentic Easy German really are for the intended audiences. As a first step in gaining insights into these issues and to further LT development for simplified German, we compiled DE-Lite, a corpus of easy-to-read texts including Easy German and comparable Standard German texts, by integrating existing collections and gathering new data from the web. We built n-gram models for an Easy German subcorpus of DE-Lite and comparable Standard German texts in order to identify typical features of Easy German. To this end, we use relative entropy (Kullback-Leibler Divergence), a standard technique for evaluating language models, which we apply here for corpus comparison. Our analysis reveals that some rules of Easy German are fairly dominant (e.g. punctuation) and that text genre has a strong effect on the distinctivity of the two language variants.

DE-Lite - a New Corpus of Easy German: Compilation, Exploration, Analysis

Large language models are prone to internalize social biases due to the characteristics of the data used for their self-supervised training scheme. Considering their recent emergence and wide availability to the general public, it is mandatory to identify and alleviate these biases to avoid perpetuating stereotypes towards underrepresented groups. We present a novel prompt-tuning method for reducing biases in encoder models such as BERT or RoBERTa. Unlike other methods, we only train a small set of additional reusable token embeddings that can be concatenated to any input sequence to reduce bias in the outputs. We particularize this method to gender bias by providing a set of templates used for training the prompts. Evaluations on two benchmarks show that our method is on par with the state of the art while having a limited impact on language modeling ability.

Prompting Fairness: Learning Prompts for Debiasing Large Language Models

This paper introduces an approach to stress identification in Tamil and Telugu, leveraging traditional machine learning models—Fasttext for Tamil and Naive Bayes for Telugu—yielding commendable results. The study highlights the scarcity of annotated data and recognizes limitations in phonetic features relevant to these languages, impacting precise information extraction. Our models achieved a macro F1 score of 0.77 for Tamil and 0.72 for Telugu with Fasttext and Naive Bayes, respectively. While the Telugu model secured the second rank in shared tasks, ongoing research is crucial to unlocking the full potential of stress identification in these languages, necessitating the exploration of additional features and advanced techniques specified in the discussions and limitations section.

Pinealai_StressIdent_LT-EDI@EACL2024: Minimal Configurations for Stress Identification in Tamil and Telugu

Stress detection from social media texts has proved to play an important role in mental health assessments. People tend to express their stress on social media more easily. Analysing and classifying these texts allows for improvements in development of recommender systems and automated mental health assessments. In this paper, a GPT model is used for classification of social media texts into two classes - stressed and not-stressed. The texts used for classification are in two Dravidian languages - Tamil and Telugu. The results, although not very good shows a promising direction of research to use GPT models for classification.

Downloads

Next from EACL 2024

Zero-shot learning for multilingual discourse relation classification

Similar lecture

Debiasing should be Good and Bad

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from EACL 2024

Zero-shot learning for multilingual discourse relation classification

Similar lecture

Debiasing should be Good and Bad

Downloads