Croatia

Pre-trained language models have been known to perpetuate biases from the underlying datasets to downstream tasks. However, these findings are predominantly based on monolingual language models for English, whereas there are few investigative studies of biases encoded in language models for languages beyond English. In this paper, we fill this gap by analysing gender bias in West Slavic language models. We introduce the first template-based dataset in Czech, Polish, and Slovak for measuring gender bias towards male, female and non-binary subjects. We complete the sentences using both mono- and multilingual language models and assess their suitability for the masked language modelling objective. Next, we measure gender bias encoded in West Slavic language models by quantifying the toxicity and genderness of the generated words. We find that these language models produce hurtful completions that depend on the subject&#39;s gender. Perhaps surprisingly, Czech, Slovak, and Polish language models produce more hurtful completions with men as subjects, which, upon inspection, we find is due to completions being related to violence, death, and sickness.

EACL 2023

Measuring Gender Bias in West Slavic Language Models

slovak

czech

polish

gender bias

Pre-trained language models have been known to perpetuate biases from the underlying datasets to downstream tasks. However, these findings are predominantly based on monolingual language models for English, whereas there are few investigative studies of biases encoded in language models for languages beyond English. In this paper, we fill this gap by analysing gender bias in West Slavic language models. We introduce the first template-based dataset in Czech, Polish, and Slovak for measuring gender bias towards male, female and non-binary subjects. We complete the sentences using both mono- and multilingual language models and assess their suitability for the masked language modelling objective. Next, we measure gender bias encoded in West Slavic language models by quantifying the toxicity and genderness of the generated words. We find that these language models produce hurtful completions that depend on the subject's gender. Perhaps surprisingly, Czech, Slovak, and Polish language models produce more hurtful completions with men as subjects, which, upon inspection, we find is due to completions being related to violence, death, and sickness.

workshop paper

### Welcome to the 17th 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 had selected the Ukrainian community for the 16th EACL, which took place online due to the COVID pandemic. Unfortunately, the ongoing war made the organisation in Kyiv impossible. Considering the importance of physical interaction among researchers, especially after the restrictions imposed by the COVID pandemic, in addition to an online mode, the conference will be held in Dubrovnik, Croatia, from 2 to 6 of May, 2023. The original aim of strengthening the connection with the Ukrainian community will not change as our program will feature a dedicated session and a workshop to highlight work on Ukrainian language technologies. 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.

You need to register for the conference in order to access this site. Please visit https://2023.eacl.org/registration for more information.

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.

This paper reports on the results of prelimi- nary experiments on the detection of persua- sion techniques in online news in Polish and Russian, using a taxonomy of 23 persuasion techniques. The evaluation addresses differ- ent aspects, namely, the granularity of the per- suasion technique category, i.e., coarse- (6 la- bels) versus fine-grained (23 labels), and the focus of the classification, i.e., at which level the labels are detected (subword, sentence, or paragraph). We compare the performance of mono- verus multi-lingual-trained state-of-the- art transformed-based models in this context.

On Experiments of Detecting Persuasion Techniques in Polish and Russian Online News: Preliminary Study

This presentation describes Adam Mickiewicz University’s (AMU) solution for the 4th Shared Task on SlavNER. The task involves the identification, categorization, and lemmatization of named entities in Slavic languages. Our approach involved exploring the use of foundation models for these tasks. In particular, we used models based on the popular BERT and T5 model architectures. Additionally, we used external datasets to further improve the quality of our models. Our solution obtained promising results, achieving high metrics scores in both tasks. We describe our approach and the results of our experiments in detail, showing that the method is effective for NER and lemmatization in Slavic languages. Additionally, our models for lemmatization will be available at: https://huggingface.co/amu-cai.

Exploring the Use of Foundation Models for Named Entity Recognition and Lemmatization Tasks in Slavic Languages

Large Language Models for Multilingual Slavic Named Entity Linking

Novel intent discovery automates the process of grouping similar messages (questions) to identify previously unknown intents. However, current research focuses on publicly available datasets which have only the question field and significantly differ from real-life datasets. This paper proposes methods to improve the intent discovery pipeline deployed in a large e-commerce platform. We show the benefit of pre-training language models on in-domain data: both self-supervised and with weak supervision. We also devise the best method to utilize the conversational structure (i.e., question and answer) of real-life datasets during fine-tuning for clustering tasks, which we call Conv. All our methods combined to fully utilize real-life datasets give up to 33pp performance boost over state-of-the-art Constrained Deep Adaptive Clustering (CDAC) (Lin et al., 2020) model for question only. By comparison CDAC model for the question data only gives only up to 13pp performance boost over the naive baseline.

Going beyond research datasets: Novel intent discovery in the industry setting

In recent years, the pattern of news consumption has been changing. The most popular multimedia news formats are now multimodal – the reader is often presented not only with a textual article but also with a short, vivid video. To draw the attention of the reader, such video-based articles are usually presented as a short textual summary paired with an image thumbnail.
In this paper, we introduce MLASK(MultimodaL Article Summarization Kit) – a new dataset of video-based news articles paired with a textual summary and a cover picture, all obtained by automatically crawling several news websites. We demonstrate how the proposed dataset can be used to model the task of multimodal summarization by training a Transformer-based neural model. We also examine the effects of pre-training when the usage of generative pre-trained language models helps to improve the model performance, but (additional) pre-training on the simpler task of text summarization yields even better results. Our experiments suggest that the benefits of pre-training and using additional modalities in the input are not orthogonal.

MLASK: Multimodal Summarization of Video-based News Articles

This article offers an overview of the IsiXhosa Intellectual Traditions Digital Archive, which hosts digitized texts and images of early isiXhosa newspapers and books from 1870-1914. The archive offers new opportunities for a range of research across multiple fields, and responds to debates around the importance of African intellectual traditions and their indigenous language sources in generating African social sciences which is contextually relevant. We outline the content and context of these materials and offer qualitative and quantitative details with the aim of providing an overview for interested scholars and a reference for those using the archive.

IsiXhosa Intellectual Traditions Digital Archive: Digitizing isiXhosa texts from 1870-1914

This paper introduces two multilingual government themed corpora in various South African languages. The corpora were collected by gathering the South African Government newspaper (Vuk'uzenzele), as well as South African government speeches (ZA-gov-multilingual), that are translated into all 11 South African official languages. The corpora can be used for a myriad of downstream NLP tasks. The corpora were created to allow researchers to study the language used in South African government publications, with a focus on understanding how South African government officials communicate with their constituents. 

In this paper we highlight the process of gathering, cleaning and making available the corpora. We create parallel sentence corpora for Neural Machine Translation (NMT) tasks using Language-Agnostic Sentence Representations (LASER) embeddings. With these aligned sentences we then provide NMT benchmarks for 9 indigenous languages by fine-tuning a massively multilingual pre-trained language model.

Preparing the Vuk'uzenzele and ZA-gov-multilingual South African multilingual corpora

This paper presents a spell checker and correction tool specifically designed for Wolof, an under-represented spoken language in Africa. The proposed spell checker leverages a combination of a trie data structure, dynamic programming, and the weighted Levenshtein distance to generate suggestions for misspelled words. We created novel linguistic resources for Wolof, such as a lexicon and a corpus of misspelled words, using a semi-automatic approach that combines manual and automatic annotation methods. Despite the limited data available for the Wolof language, the spell checker’s performance showed a predictive accuracy of 98.31% and a suggestion accuracy of 93.33%.
Our primary focus remains the revitalization and preservation of Wolof as an Indigenous and spoken language in Africa, providing our efforts to develop novel linguistic resources. This work represents a valuable contribution to the growth of computational tools and resources for the Wolof language and provides a strong foundation for future studies in the automatic spell checking and correction field.

Automatic Spell Checker and Correction for Under-represented Spoken Languages: Case Study on Wolof

We present the SpeechReporting
database, an online collection of corpora annotated for a range of discourse phenomena. The corpora contain folktales from 7 lesserstudied West African languages. Apart from its value for theoretical linguistics, especially
for the study of reported speech, the database is an important resource for the preservation of intangible cultural heritage of minority languages and the development and testing of cross-linguistically applicable computational tools.

SpeechReporting Corpus: annotated corpora of West African traditional narratives

This paper showcases new research avenues made possible by applying computational methods to historical isiXhosa text. I outline a method for isiXhosa computational text analysis which adapts word frequency analysis to be applied to isiXhosa texts focusing on root words. The paper showcases the value of the approach in a study of emerging political identities in early African nationalism, examining a novel dataset of isiXhosa newspapers from 1874 to 1890. The analysis shows how a shared identity of ‘Blackness’ (Abantsundu and Abamnyama) dynamically emerged, and follows the impact of leading intellectuals as well as African voter mobilization in shaping communal political discourse.

Analyzing political formation through historical isiXhosa text analysis: Using frequency analysis to examine emerging African Nationalism in South Africa

Measuring Gender Bias in West Slavic Language Models

Downloads

Next from EACL 2023

On Experiments of Detecting Persuasion Techniques in Polish and Russian Online News: Preliminary Study

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES