question-answering

The paper introduces horizontal and vertical proactivity in tool-using agents and proposes Q&D, a training method that rewards a questioner for retrieving required evidence without a reward model or judge. On multi-hop QA benchmarks and a simulated customer-service setting, the trained questioner beats a prompted model 15x its size on two of three benchmarks, showing that what agents pursue and when they stop matters as much as whether they act unprompted.

This paper introduces Grounded Entity Biographies (GEB), a framework for augmenting long-video memory to track entities across events, demonstrating improved performance on question-answering benchmarks.

This paper proposes a knowledge graph-based evaluation framework for assessing the contextual understanding of large language models in question answering, introducing a new similarity measure called S3KG that achieves superior performance over existing baselines.

The article proposes HiCoMER, a framework for hierarchical collaborative memory management in LLM agents that improves validity-aware retrieval, reducing outdated memories and enhancing question-answering quality.

This paper evaluates small language models beyond answer accuracy in knowledge graph question answering by isolating graph navigation capabilities, showing significant differences in path fidelity and the need for broader evaluation metrics.

UniDataAgent is an ontology-grounded agent system designed for enterprise question-to-report automation, improving efficiency and accuracy by separating semantic acquisition from online execution, as demonstrated in a deployment at China Unicom.

The paper introduces Question's Gambit, a first-move retrieval module that improves deep research agent performance by decomposing questions and providing a ranked context before iterative search, showing significant accuracy gains on BrowseComp-Plus.

This paper introduces Hallucination-R1, a robustness-oriented paraphrase generation framework designed to identify and improve factual consistency in large language models by creating semantically equivalent yet challenging paraphrases.

VisPath introduces a visual-intent-guided path reasoning framework for multimodal knowledge graph question answering, achieving significant improvements over baselines on a new benchmark, VisPath-Bench, and existing datasets.

PhysioBench introduces a unified benchmark for physiological signal question answering, harmonizing 22 datasets into 61.4 million questions across 30 tasks to evaluate the performance of various AI models.

This tweet promotes AskJev.net, a web service that allows users to ask questions and receive answers from an AI named Jev, integrated with TypeSafe's AI tools.

This paper proposes VisKG-LM, a method that compiles knowledge graphs into visual memory for efficient multiple-choice question answering, achieving performance gains over baselines by decoupling graph encoding from language reasoning.

CapMem is a human-annotated benchmark for episodic memory in egocentric video using captions, showing that caption-based QA outperforms direct video QA on long videos.

This paper benchmarks Knowledge Graph-based augmentation against Retrieval Augmented Generation for culturally specific question answering, demonstrating significant error reduction on Latin American culture data using the LatamQA dataset.

NepKANUN is an AI-powered legal assistant for Nepali using RAG and a fine-tuned LLaMA 3.2 3B model to provide accurate answers to legal queries, validated by expert reviews.

ORQA is a framework for generating occupation-realistic question-answer pairs from trusted sources to evaluate LLMs' professional knowledge, covering 116 occupations and testing 15 models, with top performers like Claude Opus 4.6 and GPT-5.4 achieving around 58-62% accuracy.

This paper details the Eloquence team's approaches for Task 2 of the Interspeech 2026 MLC-SLM challenge, which involves multilingual multiple-choice question answering using speech LLMs with fine-tuning, in-context learning, and retrieval systems.

This paper introduces the Conformal Relevance framework, which uses in-context learning and ensembling to improve conciseness in NLP tasks while maintaining coverage guarantees via conformal prediction.

APEx is a hierarchical experience utilization framework that organizes agent history into memories and skills, using a closed-loop architecture to improve deep research question answering, achieving state-of-the-art performance surpassing GPT-5.4.

The paper proposes a latent-space intervention method to improve cross-lingual factual consistency in large language models, achieving better alignment without loss in factual accuracy.