Scholarlink

Today, a crucial question in automated opinion mining is the search for consensus and patterns of its formation within public dialogue and deliberation. In our earlier works, we explored the turn of diverse individual opinions into shared meta-opinions. This paper continues the investigation of meta-opinions and examines the properties of their approximations, with particular attention to methods partially grounded in bisimulation contraction and other epistemic logical frameworks of the Kripke type. Research gap is significant complexity to find such a minimal model that is capable of representing possible worlds and the relations among them. We generated heuristics using LLM and measured metrics of this algorithm up to the fifth modal depth. The metrics shown larger than 90% accuracy even for arbitrary initial models.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Machine Learning for Meta-opinion Prediction in Social Systems Using Compact Belief Representations

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Denis N. Fedyanin,
  • Svetlana S. Bodrunova,
  • Dmitrii A. Rybakov,
  • Iuliia S. Leonova

摘要

Today, a crucial question in automated opinion mining is the search for consensus and patterns of its formation within public dialogue and deliberation. In our earlier works, we explored the turn of diverse individual opinions into shared meta-opinions. This paper continues the investigation of meta-opinions and examines the properties of their approximations, with particular attention to methods partially grounded in bisimulation contraction and other epistemic logical frameworks of the Kripke type. Research gap is significant complexity to find such a minimal model that is capable of representing possible worlds and the relations among them. We generated heuristics using LLM and measured metrics of this algorithm up to the fifth modal depth. The metrics shown larger than 90% accuracy even for arbitrary initial models.