革新知能統合研究センター

自然言語処理のトップカンファレンスの一つであるEMNLP 2025において、AIPセンターから6本の論文が採択されました。

EMNLP2025は、11月4日～9日にSuzhou（中国）で開催されます。

[EMNLP 2025]

Main Conference Oral

• Seeing is Believing, but How Much? A Comprehensive Analysis of Verbalized Calibration in Vision-Language Models
  Weihao Xuan (The University of Tokyo / RIKEN AIP)
  Qingcheng Zeng (Northwestern University, US)
  Heli Qi (RIKEN AIP / Waseda University)
  Junjue Wang (The University of Tokyo)
  Naoto Yokoya (The University of Tokyo / RIKEN AIP)

Main Conference Poster

• Bias Mitigation or Cultural Commonsense? Evaluating LLMs with a Japanese Dataset
  Taisei Yamamoto(The University of Tokyo / RIKEN AIP)
  Ryoma Kumon (The University of Tokyo / RIKEN AIP)
  Danushka Bollegala(University of Liverpool)
  Hitomi Yanaka (The University of Tokyo / RIKEN AIP)
• Thinking Out Loud: Do Reasoning Models Know When They’re Right?
  Qingcheng Zeng (Northwestern University, US)
  Weihao Xuan (The University of Tokyo/RIKEN AIP)
  Leyang Cui (Westlake University)
  Rob Voigt (University of California, Davis)

Findings

• A Unified Framework for N-ary Property Information Extraction in Materials
  Science
  Van-Thuy Phi (RIKEN AIP)
  Yuji Matsumoto (RIKEN AIP)
• Entity Profile Generation and Reasoning with LLMs for Entity Alignment
  Rumana Ferdous Munne (RIKEN AIP)
  Md Mostafizur Rahman (Rakuten Institute of Technology
  Yuji Matsumoto (RIKEN AIP)
• Post Persona Alignment for Multi-Session Dialogue Generation
  Yi-Pei Chen (University of Tokyo/RIKEN AIP)
  Noriki Nishida (RIKEN AIP)
  Hideki Nakayama (University of Tokyo)
  Yuji Matsumoto (RIKEN AIP)

関連研究室