图书情报工作 ›› 2022, Vol. 66 ›› Issue (2): 136-148.DOI: 10.13266/j.issn.0252-3116.2022.02.015

• 知识组织 • 上一篇    下一篇

一种基于知识元变异的ESI研究前沿知识演进分析方法

孙震1, 冷伏海2   

  1. 1. 山东理工大学信息管理研究院 淄博 255000;
    2. 中国科学院科技战略咨询研究院 北京 100190
  • 收稿日期:2021-07-01 修回日期:2021-09-08 出版日期:2022-01-20 发布日期:2022-02-11
  • 通讯作者: 冷伏海,战略情报研究所所长,研究员,博士,博士生导师,通信作者,E-mail:lengfuhai@casipm.ac.cn
  • 作者简介:孙震,馆员,博士。
  • 基金资助:
    本文系国家社会科学基金项目“追踪研究前沿创新要素的领域知识元方法研究”(项目编号:21CTQ025)研究成果之一。

An ESI Research Fronts Knowledge Evolution Analysis Method Based on Knowledge Element Variation

Sun Zhen1, Leng Fuhai2   

  1. 1. Institute of Information Management, Shandong University of Technology, Zibo 255000;
    2. Institutes of Science and Development, Chinese Academy of Sciences, Beijing 100190
  • Received:2021-07-01 Revised:2021-09-08 Online:2022-01-20 Published:2022-02-11

摘要: [目的/意义] 作为一类面向学科领域科技情报需求、针对全文本关键语义计量分析、旨在实现情报自动化到知识自动化实践应用的探索研究,本文基于语义标注和机器学习等技术,在前期研究从知识元共现视角探测研究前沿演进机理基础上,进一步提出一种基于知识元变异的研究前沿知识演进分析方法。[方法/过程] 利用Word2vec词嵌入模型将知识元表示为词向量,通过计算知识元向量的欧几里得距离,利用K-means聚类方法识别具有相似语义语用关联的知识元簇集,计算历时簇集内各知识元TF-IDF值,对变异后知识元重要程度的突发变化结果进行定量测度,进而挖掘ESI研究前沿演进中的知识元变异特征和规律。[结果/结论] 通过探测结果的对比检验发现,基于知识元变异的科学计量方法,不仅是对前期研究方法的补充和拓展,使得针对研究前沿内部知识运动规律的挖掘更加具体详实,更是在时间序列范畴内,能够尽早、及时探测研究前沿未来发展动向和关键情报信号的有力证据。

关键词: 知识元, 研究前沿, 机器学习, 全文本语义分析, 钙钛矿太阳能电池

Abstract: [Purpose/significance] As an exploratory research, this paper is oriented to the needs of scientific and technological information in the specialized discipline domain, and aims to realize the quantitative analysis of key semantics of the full text and the practical application shift from "information automation" to "knowledge automation". On the basis of previous studies from the perspective of knowledge element co-occurrence to explore the evolution mechanism of ESI research fronts, this paper further proposes a research front knowledge evolution analysis method based on knowledge element variation. [Method/process] Firstly, knowledge elements were represented as word vectors by word2vec word embedding model. Then, this paper calculated Euclidean distance of knowledge element vectors, and identified knowledge element clusters with similar semantic and pragmatic association by K-means clustering method. Finally, TF-IDF values of each knowledge element in the diachronic cluster were calculated. Through the quantitatively measurement of sudden changes in the importance of knowledge elements, the characteristics and rules of knowledge element variation were mined in the process of ESI research fronts evolution. [Result/conclusion] Through the comparative test of the detection results, it is found that the scientometric method based on knowledge element variation is not only a supplement and expansion of the previous research methods, but also makes the mining of the internal knowledge movement law of ESI research fronts more specific and detailed. Moreover, in the scope of time series, it is a strong evidence that the future development trend and key information signals of the ESI research fronts can be detected as soon as possible.

Key words: knowledge element, research fronts, machine learning, full-text semantic analysis, perovskite solar cell

中图分类号: