跳到正文
arXiv cs.CL· Fangxu Yu, Tao Feng, Xuying Ning, Haolin Yang, Jiaxuan You, Tianyi Zhou·· 10 小时前AI 评分38

Epistemic Compass(EpiCompass):通过可审计认知探索引导自动化科学发现

Steering Automated Scientific Discovery via Audited Epistemic Exploration

AI 导读

研究者提出即插即用框架 Epistemic Compass(EpiCompass),用认知树关联假设、证据、信念与修订,维护显式认知状态以引导自动化科学发现。该方法按预期认知增益选择候选假设,兼顾已有假设细化和假设空间扩展,并依据累积证据校验最终结论。在三个科学发现基准和三个 LLM 主干上,EpiCompass 均优于基线,并减少了无证据支撑论断与未能修正信念两类错误。

正文 · 原文

View PDF HTML (experimental)

Abstract:Scientific discovery requires revisiting the hypotheses in light of new evidence and identifying promising directions beyond those already explored. However, research agents often make claims that are not supported by their observations, explore a narrow hypothesis space, and retain beliefs despite recognizing conflicting evidence. These failures motivate the search for an explicit epistemic state that captures both beliefs about existing hypotheses and directions for further exploration. To this end, we propose Epistemic Compass (EpiCompass), a plug-and-play scaffold that maintains this state through an epistemic tree linking hypotheses, evidence, beliefs, and revisions. As an investigation progresses, EpiCompass reassesses the hypotheses explored against accumulated evidence and uses unresolved conflicts and unexplored directions to guide hypothesis generation. It selects candidates by anticipated epistemic gain, accounting for both refinement of existing hypotheses and expansion of the hypothesis space, and verifies the final conclusion against the accumulated evidence. Experimental results across three scientific discovery benchmarks and three LLM backbones demonstrate that EpiCompass consistently outperforms baselines, with further analyses showing reductions in errors of unsupported claims and failures to revise beliefs.
Comments: Technical Report
Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Cite as: arXiv:2611.25214 [cs.AI]
  (or arXiv:2611.25214v1 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2611.25214

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Fangxu Yu [view email]
[v1] Wed, 18 Nov 2026 21:39:05 UTC (798 KB)

来源:arXiv cs.CL · arxiv.org