|
|
|
| Identification of Key Core Technology Innovation Opportunities by Integrating LangChain and Multi-dimensional Technology Innovation Maps |
| Liu Peng1,2, Wei Chenyu1,2, Zhang Ke3,4,5, Zhou Wei6 |
1.School of Management, Zhengzhou University, Zhengzhou 450001
2.Henan Innovation Method Engineering Technology Research Center, Zhengzhou 450001
3.School of Information Management, Zhengzhou University, Zhengzhou 450001
4.Data Governance Research Center of Henan Province, Zhengzhou 450001
5.Research Center for “Double World-Class Project” of Henan Province, Zhengzhou 450001
6.School of Information Management, Wuhan University, Wuhan 430072 |
|
|
|
|
Abstract A comprehensive innovation breakthrough in key core technologies is a crucial foundation for building a manufacturing powerhouse and innovative nation, as well as a core driver for promoting high-quality economic development. To accelerate the pace of innovation in key core technologies, this study employed exponential random graph models (ERGM) to explore the nodal relationships among technological innovation elements in patents and conduct prompt instruction training for innovation scheme generation using large language models, thereby providing a more efficient and intelligent approach for identifying innovation opportunities in key core technologies. First, key core technologies were identified based on patent citation networks, and an evaluation system was constructed for key core technology indicators. Second, the BERTopic-ERGM model was used to mine potential combinations of innovation elements from the selected patent texts. Third, by integrating the LangChain framework with a multidimensional technological innovation map, domain-specific prompt instructions for innovative schemes were designed to enable the automatic identification of innovation opportunities in key core technologies. Furthermore, poorly generated technical solutions were analyzed in depth to further improve the overall model performance. Finally, the effectiveness of the model was validated using a dual evaluation method. The results showed that the proposed BERTopic-ERGM and LangChain-multidimensional technological innovation map model can efficiently identify innovation opportunities in the field. Compared with traditional technology forecasting methods, the proposed method not only captures emerging technological opportunities and trends more accurately, but it also enables a deeper analysis of complex interrelationships and evolutionary pathways among technologies.
|
|
Received: 24 March 2025
|
|
|
|
1 吴晓波, 邵贻玥, 林福鑫. 突破关键核心技术创新“死亡之谷”的路径机制[J]. 科学学研究, 2026, 44(1): 129-140.
2 张树满, 原长弘. 制造业领军企业如何培育关键核心技术持续创新能力?[J]. 科研管理, 2022, 43(4): 103-110.
3 李树文, 罗瑾琏, 张志菲. 从定位双星到布局寰宇: 专精特新企业如何借助关键核心技术突破实现价值共创[J]. 南开管理评论, 2024, 27(3): 94-107.
4 师磊, 阳镇, 钱贵明. 数字产业集群政策与关键核心技术突破式创新[J]. 中国工业经济, 2025(1): 100-117.
5 王昶, 彭佳琪, 何琪. 机会窗口何以触发关键核心技术创新突破?[J]. 科学学研究, 2024, 42(10): 2191-2205.
6 冯立杰, 周炜, 王金凤, 等. 基于SAO语义分析的多维技术演化路径与技术创新机会识别研究[J]. 情报学报, 2022, 41(11): 1149-1160.
7 崔宏超, 张凤, 胡京波. 创新联合体推动关键核心技术创新的价值共创机制研究[J]. 科研管理, 2025, 46(7): 81-90.
8 吕鲲, 罗星雨, 靖继鹏. 融合CFDP和CPM分析的关键核心技术识别及其路径分析——以芯片光刻领域为例[J]. 图书情报工作, 2024, 68(16): 75-89.
9 王金凤, 阎竞博, 冯立杰, 等. 知识元视域的技术元抽取及技术机会识别[J]. 情报杂志, 2024, 43(4): 96-105.
10 赵丹晓, 龚红, 董姗. 知识成熟度对关键核心技术创新的影响研究[J]. 管理学报, 2025, 22(12): 2271-2279.
11 聂力兵, 龚红, 赖秀萍. 唤醒“沉睡专利”: 知识重组时滞、重组频率与关键核心技术创新[J]. 南开管理评论, 2024, 27(8): 86-97, 160.
12 张艺, 陈凯华, 朱桂龙. 产学研合作与后发国家创新主体能力演变——以中国高铁产业为例[J]. 科学学研究, 2018, 36(10): 1896-1913.
13 张羽飞, 刘兵, 原长弘. 科技领军企业如何突破关键核心技术?——基于产学研融合视角的双案例研究[J]. 科学学与科学技术管理, 2024, 45(12): 88-105.
14 杨震宁, 袁梓晋. 数字创新网络嵌入与关键核心技术攻关[J]. 中国工业经济, 2025(5): 156-173.
15 邬爱其, 朱国浩. 数字技术对中小企业关键核心技术突破的影响——来自首台套首批次创新的证据[J/OL]. 科学学研究, (2025-05-14) [2025-07-02]. https://doi.org/10.16192/j.cnki.1003-2053.20250514.003.
16 王海花, 孙倩如, 刘黎, 等. 专精特新企业技术专业化与关键核心技术突破[J]. 科学学研究, 2026, 44(1): 141-153.
17 冯立杰, 秦浩, 张珂, 等. 基于离群专利和多维技术创新地图的技术机会识别路径研究[J]. 情报理论与实践, 2023, 46(9): 79-86.
18 冯立杰, 尤鸿宇, 王金凤. 专利技术创新路径识别及其新颖性评价研究[J]. 情报学报, 2021, 40(5): 513-522.
19 Li Y X, Yu Y F, Liang C, et al. LoftQ: LoRA-fine-tuning-aware quantization for large language models[PP/OL]. V4. arXiv (2023-11-28) [2025-07-02]. https://doi.org/10.48550/arXiv.2310.08659.
20 Kim J Y, Howard M, Cox Pahnke E, et al. Understanding network formation in strategy research: exponential random graph models[J]. Strategic Management Journal, 2016, 37(1): 22-44.
21 Dettmers T, Pagnoni A, Holtzman A, et al. QLoRA: efficient finetuning of quantized LLMs[C]// Proceedings of the 37th Advances in Neural Information Processing Systems. Red Hook: Curran Associates, 2023: 10088-10115.
22 Chavan A, Liu Z, Gupta D, et al. One-for-all: generalized LoRA for parameter-efficient fine-tuning[PP/OL]. V2. arXiv (2023-10-16) [2025-07-02]. https://doi.org/10.48550/arXiv.2306.07967.
23 Guo L, Yan F, Li T, et al. An automatic method for constructing machining process knowledge base from knowledge graph[J]. Robotics and Computer-Integrated Manufacturing, 2022, 73: 102222.
24 杨跃翔, 郑怀城, 刘学文, 等. 智能家居信息安全知识图谱构建研究[J]. 计算机科学与探索, 2026, 20(1): 169-181.
25 唐钰桦, 任晓东, 吴剑锋, 等. 联合大语言模型和跨模态时序框架的高精度电离层TEC预报方法[J/OL]. 武汉大学学报(信息科学版), (2025-06-18) [2025-07-01]. https://doi.org/10.13203/j.whugis20250104.
26 孙晓宁, 张爽, 郭文智. 知识增强大语言模型与搜索引擎用户的感知信息价值差异比较研究——基于任务复杂性与信息获取场景[J]. 图书情报工作, 2025, 69(12): 90-103.
27 Venkatasubramanian V, Chakraborty A. Quo vadis ChatGPT? From large language models to large knowledge models[J]. Computers Chemical Engineering, 2025, 192: 108895.
28 Song B W, Luan C J, Liang D N. Identification of emerging technology topics (ETTs) using BERT-based model and sematic analysis: a perspective of multiple-field characteristics of patented inventions (MFCOPIs)[J]. Scientometrics, 2023, 128(11): 5883-5904.
29 Keswani G, Bisen W, Padwad H, et al. Abstractive long text summarization using large language models[J]. International Journal of Intelligent Systems and Applications in Engineering, 2024, 12(12s): 160-168.
30 Noh H, Kim K, Song Y K, et al. Opportunity-driven technology roadmapping: the case of 5G mobile services[J]. Technological Forecasting and Social Change, 2021, 163: 120452.
31 张振刚, 罗泰晔. 基于知识组合理论的技术机会发现[J]. 科研管理, 2020, 41(8): 220-228.
32 吴柯烨, 孙建军, 张力, 等. 弱链接突变视角下的技术机会识别研究[J]. 图书情报工作, 2024, 68(10): 81-96.
33 翟东升, 刘鹤, 张杰, 等. 一种基于链路预测的技术机会挖掘方法[J]. 情报学报, 2016, 35(10): 1090-1100.
34 韦婷婷, 冯丹钰, 宋世领, 等. 技术机会发现领域专利挖掘方法研究述评[J]. 情报学报, 2023, 42(10): 1238-1250.
35 罗泰晔. 基于指数随机图模型的技术机会识别[J]. 情报学报, 2023, 42(11): 1300-1308.
36 尹航, 李云柯, 王志楠, 等. 科技差距视角下技术机会识别方法优化研究[J]. 科学学研究, 2025, 43(2): 300-310.
37 李子彪, 孙可远, 陈迪, 等. 基于科技差距评估GTM专利地图识别的技术机会[J]. 情报杂志, 2023, 42(7): 147-153, 44.
38 刘志迎, 姚嘉珍. 生物反应器关键核心技术识别与国别差距研究[J]. 科学学研究, 2025, 43(4): 723-732, 786.
39 孙彩虹, 蔡鸿宇, 石进, 等. 专利应用视角下的关键核心技术识别研究[J]. 情报理论与实践, 2025, 48(1): 121-130.
40 王慧扬, 刘建华, 赵玉冰. 知识搜索与科技领军企业关键核心技术突破[J/OL]. 科学学研究, (2025-06-20) [2025-07-06]. https://doi.org/10.16192/j.cnki.1003-2053.20250617.001.
41 李晓蓉, 张津明, 刘红琴. 数字化转型对中国制造业企业创新的影响研究[J]. 科研管理, 2025, 46(5): 1-12.
42 冯翠翠, 易明, 莫富传. 基于指数随机图模型的在线健康社区多元主体交互影响机制研究[J]. 图书情报工作, 2024, 68(7): 88-101.
43 郭建杰, 谢富纪. 基于ERGM的协同创新网络形成影响因素实证研究[J]. 管理学报, 2021, 18(1): 91-98.
44 朱丹浩, 黄肖宇, 李堯霖, 等. 基于大语言模型的法律文本的自动摘要方法[J]. 数据分析与知识发现, 2025, 9(6): 35-46.
45 周青, 邹凡, 何铮. 生命周期视角下产业技术创新战略联盟冲突影响因素演变研究[J]. 科技进步与对策, 2017, 34(4): 66-71.
46 吴柯烨, 孙建军, 谢紫悦. 基于专利文本挖掘的细粒度技术机会分析[J]. 情报学报, 2023, 42(10): 1199-1212.
47 任云坤, 陈思, 秦飞. 温度循环对硅通孔绝缘层漏电机制的影响[J]. 物理学报, 2025, 74(5): 165-172.
48 肖江, 吴恩平, 张世桀, 等. ErlangShen: 基于流水线低访问开销的图式区块链高效事务执行机制[J]. 电子与信息学报, 2024, 46(5): 2111-2121.
49 张芊帆, 何茜, 田雨, 等. 3D IC封装技术中硅通孔研究进展综述[J]. 电子与信息学报, 2025, 47(9): 3057-3069. |
|
|
|
