Newest papers

CrisiLangs: A Multilingual Resource for Crisis Warning Generation

Giacomo Gonella, Stefano Menini, Marco Guerini · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.31356 · cited by 0 works (OpenAlex) · open-access PDF

Portuguese

Comunicar com o público durante uma emergência envolve emitir mensagens de alerta rapidamente. Na Europa, muitas vezes é necessário produzi-las em mais de um idioma. O trabalho recente mostrou que os modelos de idioma podem ser treinados para gerar mensagens de alerta quando uma crise se desenvolve, mas os dados disponíveis cobrem apenas o inglês. Para preencher essa lacuna, apresentamos CrisiLangs, uma extensão multilingue do CrisiText (um conjunto de dados de eventos de crise em inglês emparelhado com mensagens de alerta escritas de acordo com as diretrizes de comunicação de especialistas) para o italiano, búlgaro, alemão e grego. Os falantes nativos especialistas traduziram as mensagens de alerta de 1.000 eventos extraídos de 411 cenários, para um total de 5.000 pares de mensagens de eventos em cinco

Machine translation (Meta M2M100); the English original follows.

Abstract

Communicating with the public during an emergency entails issuing warning messages promptly. In Europe, it often requires producing them in more than one language. Recent work has shown that language models can be trained to generate warning messages as a crisis unfolds, but the available data covers only English. To fill this gap, we present CrisiLangs, a multilingual extension of CrisiText (an English dataset of crisis events paired with warning messages written according to expert communication guidelines) to Italian, Bulgarian, German, and Greek. Expert native speakers translated the warning messages of 1,000 events drawn from 411 scenarios, for a total of 5,000 event-message pairs across five languages. We test the resource to fine-tune a multilingual model on all five languages jointly, with and without the emergency guidelines in the prompt, and compare it with a zero-shot baseline. The fine-tuned models outperform the baseline on ROUGE-L, BLEU and BERTScore in every language and configuration. In a human preference study, low decoding temperatures receive the most preferences in all five languages and in both configurations, while the effect of the guidelines varies by language.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Resource (disambiguation), Warning system, Preference, Natural language processing, Language model

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange