| 黄悦悦 张敏 李新丹 张亚婕.AI 赋能下的科技查新范式演进: 基于多维绩效的实证研究[J].中国科技资源导刊,2026,(3):45~51 |
| AI 赋能下的科技查新范式演进: 基于多维绩效的实证研究 |
| Evolution of Technology Novelty Search Paradigms Empowered by AI: An Empirical Study Based on Multi-dimensional Performance |
| 投稿时间：2026-01-15 |
| DOI： |
| 中文关键词: 科技查新;大语言模型;人机协同;范式演进;实证研究 |
| 英文关键词: scientific and technological novelty search, large language models (LLMs), human-AI collaboration, paradigm evolution, empirical research |
| 基金项目:河南省科学技术情报中心基本科研业务费项目“AI 赋能河南省科技情报服务优化路径研究”(202403)，河南省科学技 术情报中心基本科研业务费项目“AI 技术下科技查新精准服务提升及智慧平台建设研究”(JY202503)。 |
| 作者 | 单位 | | 黄悦悦 张敏 李新丹 张亚婕 | (河南省科学技术情报中心，河南郑州 450000) |
|
| 摘要点击次数: 31 |
| 全文下载次数: 16 |
| 中文摘要: |
| 随着大语言模型(LLMs)深度融入科技情报领域，科技查新正加速由人工主导向智能驱动转型。为厘清智能化路径的可行性与局限，需通过实证研究，系统评估传统查新、人机协同与智能主导这3种典型范式的绩效。以某省级科技情报机构的60个真实科技查新项目为样本，构建多维评价指标体系，采用双因素重复测量方差分析与TukeyHSD事后检验，并对上述3种查新范式的绩效表现进行实证比较。研究显示:人机协同范式已成为当前科技查新智能化转型的主流路径，其综合性能优势与创新稳健性彰显了人类专业判断与机器高效处理的互补价值;智能主导范式的落地应用依赖于幻觉抑制等关键技术突破，现阶段更适宜作为辅助工具;传统查新范式仍是高敏感度项目的稳妥选择。建议未来应坚持以人机协同为核心范式，依据任务敏感度与创新程度实现3类范式的动态适配。 |
| 英文摘要: |
| With the deep integration of large language models (LLMs) into the field of scientific and technological intelligence, novelty search is rapidly shifting from a human-dominated paradigm to an AI- driven one. To clarify the feasibility and limitations of this intelligent transformation, an empirical assessment is urgently needed to systematically evaluate the effectiveness of three typical paradigms: traditional novelty search, human-AI collaboration, and AI-driven approaches.A multi-dimensional evaluation index system was constructed based on 60 real-world novelty search projects from a provincial institute of scientific and technical information. Two-way repeated-measures analysis of variance (ANOVA) and Tukey’s HSD post-hoc test were used to empirically compare the performance of the aforementioned three paradigms. The research shows that the human-AI collaboration paradigm has become the mainstream approach for the intelligent transformation of current scientific and technological novelty search. Its comprehensive performance advantages and innovation robustness highlight the complementary value of professional human judgment and efficient machine processing. The practical application of the AI-driven paradigm relies on breakthroughs in key technologies such as hallucination suppression, and it is more suitable as an auxiliary tool at the current stage. The traditional novelty search paradigm remains a reliable choice for high-sensitivity projects. In the future, we should adhere to the human-AI collaboration paradigm as the core, and realize the dynamic adaptation of the three paradigms according to task sensitivity and innovation level. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
