图书情报工作 ›› 2022, Vol. 66 ›› Issue (10): 45-58.DOI: 10.13266/j.issn.0252-3116.2022.10.004

所属专题: 创新驱动战略下的技术预测方法与实践

• 专题:创新驱动战略下的技术预测方法与实践 • 上一篇    下一篇

动态主题网络视角下的突破性创新主题识别:以区块链领域为例

陈虹枢, 宋亚慧, 金茜茜, 汪雪锋   

  1. 北京理工大学管理与经济学院 北京 100081
  • 收稿日期:2021-11-21 修回日期:2022-02-16 出版日期:2022-05-20 发布日期:2022-06-01
  • 作者简介:陈虹枢,助理教授,博士,E-mail:Hongshu.Chen@bit.edu.cn;宋亚慧,硕士研究生;金茜茜,博士研究生;汪雪锋,教授,博士。
  • 基金资助:
    本文系 国家自然科学基金青年项目"多源异质网络视角下产学研合作产生机理及潜在机会发现研究"(项目编号:72004009)和北京理工大学优秀青年教师学术启动项目"基于主题模型及深度学习的技术演化路径识别研究"研究成果之一。

Radical Innovative Topic Identification from a Perspective of Dynamic Topic Network:Taking the Field of Blockchain as an Example

Chen Hongshu, Song Yahui, Jin Qianqian, Wang Xuefeng   

  1. School of Management and Economics, Beijing Institute of Technology, Beijing 100081
  • Received:2021-11-21 Revised:2022-02-16 Online:2022-05-20 Published:2022-06-01

摘要: [目的/意义]突破性创新对科技发展具有关键作用。大数据环境下,科学技术发展本身所具有的复杂、多维、不断进化等特征越发凸显。以动态视角进行突破性创新主题识别,对于为国家、企业及高校详析突破性创新领域、合理配置创新资源以及提供创新升级解决方案具有重要意义。[方法/过程] 综合运用主题模型、词嵌入算法以及复杂网络分析等方法构建动态主题网络,全面考量主题在时间窗口内的结构特性以及时间窗口间的演化状态,并以其为基础结合突破性创新的新颖性、突变性、影响力和学科交叉性特征识别突破性创新主题。[结果/结论] 面向区块链领域展开实证研究,识别出神经网络(Neural Network)和边缘计算(Edge Computing)两个主题的突破性创新特征最为显著。结合区块链现有研究及美国国家科学技术委员会发布的关键和新兴技术清单,验证了本文方法的可行性和有效性。但有关结果的定量验证,以及融合多源数据的突破性创新主题识别有待进一步研究。

关键词: 突破性创新, 主题网络, 主题识别, LDA, Word2vec模型, 区块链

Abstract: [Purpose/Significance]Radical innovation plays a key role in the development of science and technology. In the big data environment, the complex, multidimensional, and continuous evolutionary characteristics of science and technology development itself is becoming more observable than ever before. It is important to identify these topics from a dynamic perspective to provide solutions for countries, enterprises and universities to analyze radical innovation areas, allocate innovation resources rationally and seek innovation upgrades.[Method/Process] This paper integrated methods of topic modeling, word embedding algorithm, and complex network analysis to construct dynamic topic networks, and evaluate the structural characteristics of the topics within different time windows and the topic evolution states between these time windows. Based on dynamic topic networks, this paper then combined the novelty, mutation, impact and interdisciplinary characteristics of radical innovation to identify topics of radical innovation.[Result/Conclusion] Through the empirical study on blockchain, this paper recognizes that two topics with the most significant radical innovative characteristics are Neural Network and Edge Computing. With existing research of blockchain and the list of critical and emerging technologies issued by the National Science and Technology Council (NSTC) of the United States, this paper finally verifies the feasibility and effectiveness of the proposed method. However, further quantitative verification of the result of this paper, and identification of radical innovative topics by fusing multi-source data, require further research in the future.

Key words: radical innovation, topic network, topic identification, LDA, Word2vec, blockchain

中图分类号: