In the digital era, social media has emerged as a powerful channel for expressing opinions, but online platforms have also become a breeding ground for hate speech targeting individuals based on color, caste, gender, sexual orientation, and political ideologies. Despite growing interest in automatic hate speech detection, existing research remains predominantly focused on English, underscoring a critical need to extend efforts to under-resourced languages. To bridge this gap, the HASOC (Hate Speech and Offensive Content Identification) shared task has been promoting multilingual hate speech research. In this paper, we present a brief overview of these four shared tasks (Assamese, Bengali, Bodo and English), datasets, participating systems, and their performance across standard evaluation metrics—precision, recall, accuracy, and macro F1 score. In addition, we analyze the inter-system agreement using Cohen’s κ and Fleiss’ κ , and investigate item-level difficulty through hardness analyses. Our findings offer valuable insights into the challenges and progress in multilingual hate speech detection, particularly for low-resource languages. This paper also serves as a model for the analysis of other results of large-scale experimentation with text classification systems. • Hate Speech Detection for social media in low-resource Languages and English is analyzed based on experiments for 4 languages. • Approaches for Hate Speech Detection and their performance are compared. • Item hardness of social media posts is analyzed and patterns for low-resource languages are compared. • Inter-system agreement or similarity is analyzed using Cohen’s κ and Fleiss’ κ .

Abstract Chain-of-thought (CoT) traces have been shown to improve performance of large language models on a plethora of reasoning tasks, yet there is no consensus on the mechanism by which this boost is achieved. To shed more light on this, we introduce Causal CoT Graphs (CCGraphs), which are directed acyclic graphs automatically extracted from reasoning traces that model finegrained causal dependencies in language-model outputs. A collection of 1671 mathematical reasoning problems from MATH500, GSM8K, and AIME, together with their associated CCGraphs, has been compiled into our dataset—KisMATH. Our detailed empirical analysis with 15 open-weight LLMs shows that (i) reasoning nodes in the CCGraphs are causal contributors to the final answer, which we argue is constitutive of reasoning; and (ii) LLMs emphasize the reasoning paths captured by the CCGraphs, indicating that the models internally realize structures similar to our graphs. KisMATH enables controlled, graph-aligned interventions and opens avenues for further investigation into the role of CoT in LLM reasoning.

Samuel Cahyawijaya, Holy Lovenia, Joel Ruben Antony Moniz, Tack Hwa Wong, Mohammad Rifqi Farhansyah, Thant Thiri Maung, Frederikus Hudi, David Anugraha, Muhammad Ravi Shulthan Habibi, Muhammad Reza Qorib, Amit Agarwal, Joseph Marvin Imperial, Hitesh Laxmichand Patel, Vicky Feliren, Bahrul Ilmi Nasution, Manuel Antonio Rufino, Genta Indra Winata, Rian Adam Rajagede, Carlos Rafael Catalan, Mohamed Fazli Mohamed Imam, Priyaranjan Pattnayak, Salsabila Zahirah Pranida, Kevin Pratama, Yeshil Bangera, Adisai Na-Thalang, Patricia Nicole Monderin, Yueqi Song, Christian Simon, Lynnette Hui Xian Ng, Richardy Lobo Sapan, Taki Hasan Rafi, Bin Wang, Supryadi, Kanyakorn Veerakanjana, Piyalitt Ittichaiwong, Matthew Theodore Roque, Karissa Vincentio, Takdanai Kreangphet, Phakphum Artkaew, Kadek Hendrawan Palgunadi, Yanzhi Yu, Rochana Prih Hastuti, William Nixon, Mithil Bangera, Adrian Xuan Wei Lim, Aye Hninn Khine, Hanif Muhammad Zhafran, Teddy Ferdinan, Audra Aurora Izzani, Ayushman Singh, Evan Evan, Jauza Akbar Krito, Michael Anugraha, Fenal Ashokbhai Ilasariya, Haochen Li, John Amadeo Daniswara, Filbert Aurelian Tjiaranata, Eryawan Presma Yulianrifat, Can Udomcharoenchaikit, Fadil Risdian Ansori, Mahardika Krisna Ihsani, Giang Nguyen, Anab Maulana Barik, Dan John Velasco, Rifo Ahmad Genadi, Saptarshi Saha, Chengwei Wei, Isaiah Edri W. Flores, Kenneth Chen Ko Han, Anjela Gail D. Santos, Wan Shen Lim, Kaung Si Phyo, Tim Santos, Meisyarah Dwiastuti, Jiayun Luo, Jan Christian Blaise Cruz, Ming Shan Hee, Ikhlasul Akmal Hanif, M.Alif Al Hakim, Muhammad Rizky Sya’ban, Kun Kerdthaisong, Lester James Validad Miranda, Fajri Koto, Tirana Noor Fatyanosa, Alham Fikri Aji, Jostin Jerico Rosal, Jun Kevin, Robert Wijaya, Onno P. Kampman, Ruochen Zhang, Börje F. Karlsson, Peerat Limkonchotiwat. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2025.

Soumadeep Saha, Sutanoya Chakraborty, Saptarshi Saha, Utpal Garain. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Most counterfactual inference frameworks traditionally assume acyclic structural causal models (SCMs), i.e. directed acyclic graphs (DAGs). However, many real-world systems (e.g. biological systems) contain feedback loops or cyclic dependencies that violate acyclicity. In this work, we study counterfactual inference in cyclic SCMs under shift–scale interventions, i.e., soft, policy-style changes that rescale and/or shift a variable’s mechanism.

Deep learning has advanced rapidly, but relies heavily on large-labeled datasets for effective training. This is particularly challenging in fields like medicine, where expert labeling is costly, labor-intensive, and prone to bias and error. Semi-supervised learning (SSL) addresses this challenge by reducing reliance on labeled data. SSL is closely tied to the concept of causation. However, recent works relating causality to SSL are limited by modeling only low-dimensional observations or designing a plug-in module to alleviate the class imbalance. In this paper, we take steps towards training causal generative models for semi-supervised learning, combining principles from causality and variational inference. We interpret the Mixup strategy as a stochastic intervention and introduce a consistency loss to promote coherent latent representations. Under reasonable assumptions, we provide theoretical guarantees that the learned latent representations align with true causal factors up to permissible ambiguities. The experimental results show the proposed approach achieves state-of-the-art performance on several medical datasets of different modalities. Additionally, we test our model on standard benchmarking datasets: CIFAR10, CIFAR100, and SVHN, where it achieves competitive performance.