Researchers at the Seventh Workshop on Natural Language Processing and Computational Social Science found that emojis, while rarely offensive alone, can acquire harmful meanings through symbolic assoc…
Researchers Alexandria Leto and Maria Leonor Pacheco introduced an unsupervised LLM-based method for conceptual metaphor discovery in political discourse, demonstrating its utility through a case stud…
Even frontier large language models such as GPT-5.2 and leading open-weight alternatives consistently underperform fine-tuned BERT on a challenging legal reasoning task, according to a study by Caroli…
Researchers from KIT submitted a cross-lingual voice cloning system to the IWSLT 2026 track, building on FishAudio-S2-Pro with language tag prompting, reinforcement learning fine-tuning, and reference…
The CUHKSZ team submitted a simultaneous speech translation system to IWSLT 2026, built on Qwen3-Omni-30B-A3B with LoRA adaptation, achieving 40.5 BLEU for English→Chinese and 27.7 BLEU for English→Ge…
Researchers from IIIT-BGP presented low-resource Bhojpuri-Hindi speech translation systems at IWSLT 2026, including an end-to-end model combining Wav2Vec2 and NLLB-200 with a lightweight adapter, and …
Researchers from an undisclosed institution introduced AlignAtt4LLM, the first application of the AlignAtt policy to decoder-only large language models for simultaneous speech translation at IWSLT 202…
Researchers presented AURA-ST, a three-stage modular pipeline for low-resource speech-to-text translation at IWSLT 2026. The system bypasses traditional cross-attention by treating projected acoustic …
Researchers evaluated four state-of-the-art zero-shot cross-lingual voice cloning systems across six languages, finding that Arabic remains particularly challenging. The study, presented at IWSLT 2026…
Barcelona Supercomputing Center submitted an end-to-end system combining a speech encoder with a translation-oriented large language model to the instruction following track of IWSLT 2026. The system …
Researchers from the Karlsruhe Institute of Technology developed Diet-KIT, a post-training quantization system that compresses the Qwen2-Audio-7B speech LLM from 16 GB to 3.98 GB while maintaining tra…
HW-TSC submitted a cascaded system to the IWSLT 2026 Subtitling track, using a large-model-based streaming speech recognition framework with VAD, sliding-window context caching, long audio chunking, a…
Researchers at Ohio State University developed HydraQE, an end-to-end, reference-free quality estimation system for speech translation that outperforms cascaded text-based baselines. Submitted to the …
Researchers introduced Patent-CR, the first English dataset for patent claim revision, containing rejected and granted patent claims. Evaluating large language models, they found GPT-4 performed best …
Researchers at the 2025 Conference on Empirical Methods in Natural Language Processing introduced PatentScore, a multi-dimensional evaluation framework for LLM-generated patent claims. PatentScore int…
Researchers from multiple institutions published a comprehensive survey on Process Reward Models (PRMs) at ACL 2026, covering data generation, model construction, and usage for step-level reasoning ev…
Researchers introduced CoLabScience, a proactive AI assistant that uses a reinforcement learning framework called PULI to autonomously intervene in biomedical research discussions. The system, trained…
Researchers at ACL 2026 revealed PriceBlind, a visual adversarial attack that exploits a vulnerability called Visual Dominance Hallucination in multimodal large language models used as financial agent…
Researchers introduced TRuST, a large-scale dataset with ~300k annotations for toxicity recognition, unifying prior resources through a synthesized definition. Benchmarking showed fine-tuned PLMs outp…
Researchers at the Association for Computational Linguistics found that large language models' numerical accuracy drops significantly when numbers are presented in underrepresented scripts or formats,…