InternReviewer 與 InternAdvocate：同儕審查與答辯中代理式強化學習之客觀獎勵與評估

InternReviewer & InternAdvocate: Objective Reward and Evaluation for Agentic Reinforcement Learning in Peer Review and Rebuttal

July 21, 2026

作者: Xuerui Su, Liya Guo, Qizhi Pei, Qipeng Guo, Zhongbo Tian, Lijun Wu, Kai Chen, Zun Wang

cs.AI

摘要

生成專業學術內容，如同儕評審與答辯，需將領域推理與事實依據進行縝密的協同整合。本研究提出一套全面性框架，用於開發與評估專業化學術智能體——InternReviewer 與 InternAdvocate。我們首先建立一個大規模、高品質的學術資料集，並整合高效率的 arXiv 檢索工具，以實現主動式證據蒐集。為最佳化這些智能體，我們實作了一種以統一目標指標與獎勵系統驅動的智能體強化學習（RL）範式。該系統透過採用多維度評判標準，避免基於主觀模型評判的偏差，其標準包括以參考文獻為錨定的語義對齊、結構合規性，以及一套嚴格的驗證機制——該機制將引用與即時互動日誌進行交叉比對，以消除幻覺現象。實驗結果表明，在此閉環框架內訓練的智能體，在推理深度與引用準確性方面均展現出顯著提升。

English

Generating professional scholarly content, such as peer reviews and rebuttals, requires an intricate synergy between domain reasoning and factual grounding. This work presents a comprehensive framework for the development and evaluation of specialized scholarly agents, InternReviewer and InternAdvocate. We first establish a large-scale, high-quality scholarly dataset and integrate a high-efficiency arXiv retrieval tool to enable active evidence gathering. To optimize these agents, we implement an agentic Reinforcement Learning (RL) paradigm driven by a unified objective metric and reward system. This system avoids the biases of subjective model-based judging by employing multi-dimensional criteria, including reference-anchored semantic alignment, structural compliance, and a strict verification mechanism that cross-checks citations against real-time interaction logs to eliminate hallucinations. Experimental results demonstrate that agents trained within this closed-loop framework exhibit significant improvements in reasoning depth and citation accuracy.