[2604.06826] Environmental, Social and Governance Sentiment Analysis on Slovene News: A Novel Dataset and Models

Computer Science > Computation and Language arXiv:2604.06826 (cs) [Submitted on 8 Apr 2026] Title:Environmental, Social and Governance Sentiment Analysis on Slovene News: A Novel Dataset and Models Authors:Paula Dodig, Boshko Koloski, Katarina Sitar Šuštar, Senja Pollak, Matthew Purver View a PDF of the paper titled Environmental, Social and Governance Sentiment Analysis on Slovene News: A Novel Dataset and Models, by Paula Dodig and Boshko Koloski and Katarina Sitar \v{S}u\v{s}tar and Senja Pollak and Matthew Purver View PDF HTML (experimental) Abstract:Environmental, Social, and Governance (ESG) considerations are increasingly integral to assessing corporate performance, reputation, and long-term sustainability. Yet, reliable ESG ratings remain limited for smaller companies and emerging markets. We introduce the first publicly available Slovene ESG sentiment dataset and a suite of models for automatic ESG sentiment detection. The dataset, derived from the MaCoCu Slovene news collection, combines large language model (LLM)-assisted filtering with human annotation of company-related ESG content. We evaluate the performance of monolingual (SloBERTa) and multilingual (XLM-R) models, embedding-based classifiers (TabPFN), hierarchical ensemble architectures, and large language models. Results show that LLMs achieve the strongest performance on Environmental (Gemma3-27B, F1-macro: 0.61) and Social aspects (gpt-oss 20B, F1-macro: 0.45), while fine-tuned SloBERTa is the best mode...