融合Word2Vec的半积累引用共词网络的领域主题演化研究

Exploring the Evolution of Research Topics Based on Partial-Cumulative Citation Co-Word Network with Word2Vec

发现并厘清学科及领域的发展路径和演化脉络,对科学研究和学科发展具有重要意义。针对传统共词分析法的不足和主题演化分析维度的单一性问题,本文提出了一种基于引用共词网络的主题发现与演化分析方法,并以情报学领域为例进行了实证研究。通过引用关系定义“引用共现”关联,并融合词嵌入技术构建关键词网络;使用社区探测法识别领域主题,采用后离散分析法,从内容结构和发展趋势两个角度进行学科主题演化分析,并可视化呈现主题演化路径及发展趋势。研究结果表明,本文所构建网络比传统共词网络能呈现粒度更优的主题聚类效果,并且能较好地呈现主题动态演化趋势,是共词分析法的有效补充。

Discovering and clarifying the development and evolution of disciplines contribute significantly to scientific research and academic development.Targeting the single-dimension issue in regular co-word analysis,this paper proposes a subject identification and evolution analysis method based on the citation co-word network,and considers the field of“information science”as an example for empirical research.This method defines the“citation co-occurrence”association through the citation relationship,and builds a keyword network with word embedding method.Subsequently,the community detection strategy is adopted to identify field subjects,while the second discrete analysis methods are also employed to analyze the structure and trend of the subjects.The subject evolution path and trend are visualized,the results showing that the network proposed presents better topic clustering and trending visualization effect than the regular co-word networks.Indeed,the aforementioned proves that this method is an effective supplement to co-word analysis.