Natural Language Processing

Researchers introduced Outcome Reward Models (ORMs) for Text-to-SQL verification, using semantic scoring to outperform traditional heuristics. Their GradeSQL framework automates candidate generation and labeling, achievi…

Researchers testing GPT-2 style models on 'impossible' languages found that while the models showed gradual decline in grammatical sensitivity, they struggled significantly with generative production, failing to produce …

Researchers propose DDIAgents, a mechanism-conditioned multi-agent framework for drug-drug interaction prediction that dynamically orchestrates specialized expert agents and adapts context flow to the inferred interactio…

Researchers introduced a controlled student-teacher protocol to study when natural-language feedback improves multi-turn language agent performance beyond repeated attempts. Testing thirteen open-weight models across fou…

Researchers introduced a semantic-layer-mediated NL2SQL agent that decouples semantic intent from physical SQL execution, achieving 94.15% execution accuracy on the Spider2-snow benchmark and ranking third on the officia…

Researchers introduced Explanation Quality Markers (EQMs), a set of sixty reasoning patterns scored by large language models, to measure judgment quality in natural-language explanations. In a pre-registered analysis of …

Researchers submitted a system to SemEval-2026 Task 9 for detecting multilingual online polarization using transformer-based models with class weighting and threshold tuning. Their approach achieved competitive F1 macro …

Researchers deployed an automated pipeline to optimize skill descriptions for an enterprise AI agent, achieving 79.2% F1 accuracy versus 79.4% for manual tuning while reducing engineering effort per skill from 120 to 3.8…

A new study from arXiv investigates how AI can retrieve simulation models using natural language queries, finding that data representation, open-source embedding models, and reranking strategies significantly impact retr…

Researchers propose a reference-based evaluation protocol for prosody and rhythm in speech-to-speech AI agents, using 4000+ hours of dyadic English conversation to create matched reference regimes for metrics like F0 and…

Researchers introduced GradeSQL, a framework for training Outcome Reward Models (ORMs) to improve test-time verification in Text-to-SQL tasks. ORM-based selection outperformed execution-based Best-of-N and Majority Votin…

Researchers introduced a framework using generative AI agents to automate black-box audits of personalization algorithms, deploying 1,120 agents on X after the 2024 U.S. election. They found that X's algorithmic feed amp…

Researchers introduced Indi-RomCoM, a benchmark for evaluating large language models on Romanized code-mixed Indic-English instructions. Testing across seven tasks and four languages showed that LLMs consistently underpe…

Researchers introduced Contrastive Reflection, an iterative prompt-optimization framework for agentic information retrieval workflows, which uses error-anchored behavioral slices and contrastive examples to propose targe…

Researchers released a benchmark for Arabic-Russian scientific translation, including a 27,000-sentence parallel corpus and fine-tuned multilingual models. The Qwen2.5-7B model achieved BLEU 23.15, outperforming zero-sho…

A new study on natural-language-to-Lean formalization reveals that compile-rate evaluation overestimates faithfulness, with a tool-augmented agent achieving 89.5% compilation but only 60.5% consensus faithfulness. The re…

Researchers extend Frictive Policy Optimization (FPO) to handle perceptual asymmetry in dialogue, where participants have asymmetric partial information. Cross-corpora analysis and LLM probing on the HCRC MapTask show th…

Researchers argue that AI agents should help users construct preferences rather than assume they have well-formed ones, introducing CoPref and CoShop to model and benchmark this. Testing five frontier models on CoShop, n…

Researchers introduced a Bangla event detection benchmark with 9,979 annotated sentences across clean text, ASR transcripts, and corrupted text. They found encoder models perform better on clean text but degrade under no…

Researchers propose CORTEX, a token-level hallucination detection method for Retrieval-Augmented Generation (RAG) that identifies ungrounded content by comparing internal representations of a large language model with an…