Under review 2026
ConSieve uses counterfactual supervision to select evidence that satisfies answer-critical query constraints during RAG context compression.
Under review 2026
ConSieve uses counterfactual supervision to select evidence that satisfies answer-critical query constraints during RAG context compression.
Under review 2026
This study examines how human-rated and language-derived forms of novelty relate to citations and public attention in NLP and machine learning.
Under review 2026
This study examines how human-rated and language-derived forms of novelty relate to citations and public attention in NLP and machine learning.
Under review 2026
AgentPersonaBench evaluates whether latent persona traits guide agents' observable behavior across survey, chat, web, and desktop application tasks.
Under review 2026
AgentPersonaBench evaluates whether latent persona traits guide agents' observable behavior across survey, chat, web, and desktop application tasks.
Under review 2026
CRED combines enforced methodological diversity with live competition between research agents to improve automated research under limited time and evaluation budgets.
Under review 2026
CRED combines enforced methodological diversity with live competition between research agents to improve automated research under limited time and evaluation budgets.

Furui Xu, Shaobo Wang, Jiajun Zhang, Chenghao Sun, Haixiang Tang, Linfeng Zhang
AAAI 2026 Poster
We conduct dataset pruning from the perspective of generalization.
Furui Xu, Shaobo Wang, Jiajun Zhang, Chenghao Sun, Haixiang Tang, Linfeng Zhang
AAAI 2026 Poster
We conduct dataset pruning from the perspective of generalization.