“四学派”框架下GPT技术在中文论文同行评审中的应用与展望——以中文图情档论文评审GPTs为例  

Application and Prospects of GPT Technology in Peer Review of Chinese Academic Papers Under the“Four Schools”Framework:Case Study of GPTs for Reviewing Chinese Library and Information Science Papers

作　　者：许舒婷 Xu Shuting(Party School of the Fujian Provincial Committee of CPC(Fujian Academy of Governance),Fuzhou Fujian 350108)

机构地区：[1]中共福建省委党校福建行政学院,福建福州350108

出　　处：《情报探索》2025年第2期84-93,共10页Information Research

摘　　要：[目的/意义]GPT技术为同行评审带来新的应用场景、模式融合和伦理难题。同行评审“四学派”框架为GPT技术在中文论文同行评审中的应用提供了设计思路、评估指标和改进方向。[方法/过程]以中文图情档论文评审GPTs为例,设计同行评审GPTs的通用流程,创建专业知识库,运用CO-STAR提示词框架,通过文本相似度分析、统计分析、重叠度指标、伦理对话测试等方法对评审效果展开实测。[结果/结论]GPTs生成的论文总结与原文摘要文本相似度高于80%,具有读懂学术论文的能力。GPTs能够稳定执行评审流程,其评奖等级具有参考价值,并能坚持学术伦理。尽管存在评分较高、评分受顺序影响、对齐程度不同等问题,其仍具有平衡“四学派”各目标的良好执行效果。s exceeds 80%,demonstrating the ability to understand academic papers.GPTs can reliably execute the review process,with evaluation results being of reference value and maintaining academic ethics.Although there are some problems such as giving high score,the score being affected by the order,and different degree of alignment,GPTs still has a good execution effect of balancing the goals of the“Four Schools”.

关 键 词：四学派 同行评审 GPT技术 GPTs 

分 类 号：G353[文化科学—情报学]