慢病知识网络的层级结构与内部关联方法研究——以高血压为例

Hierarchical Structure and Internal Correlation Method of the Chronic Disease Knowledge Network:Taking Hypertension as an Example

[目的/意义]探寻慢病知识网络的层级结构与内部关联对深入认识慢病知识网络、发挥图情学科在慢病知识组织方面的优势具有重要意义。[方法/过程]以Pub Med数据库中下载的26717篇与高血压相关的文献题录数据为研究对象构建慢病知识网络,提出一种描述性的剖析网络横截面的层析内容分析方法 (TCA),以K-core值和聚类系数为指标对知识网络进行层级划分,从定量的角度比较网络各层级之间的关联和差异;同时,将网络分层的结果与通过传统社会网络分析得到的各社群模块进行相似度对比分析。[结果/结论]慢病知识网络可被划分为基础层、中间层和细节层,各层之间差异比较明显,基础层可用于划分网络的主要研究方向,细节层可用于揭示网络的微观形态,而中间层可用于探索网络的演化路径;与传统的社会网络分析方法相比,该方法能够深入挖掘知识网络中的层级结构与内部关联形态。

[Purpose/significance] Exploring the hierarchical structure and internal correlation of the chronic disease knowledge network is of great significance to deeply understand chronic disease knowledge network and give play the disciplinary advantage of the Library and Information Science in chronic disease. [Method/process] In this paper,bibliographic data of 26717 articles related to hypertension downloaded from Pub Med are used to construct the knowledge network of chronic disease,and a descriptive Tomographic Content Analysis method is proposed to analyze the cross-section of the network. The K-core values and clustering coefficients are applied to classify the knowledge network layers and compare their correlations and differences from a quantitative point of view. At the same time,the results of network stratification are compared with the community modules obtained by the traditional social network analysis. [Result/conclusion] The research shows that the knowledge network of chronic disease can be divided into the basic layer,the middle layer and the detail layer,and the differences among them are obvious. The basic layer can be used to divide the main research directions,the detail layer can be used to reveal the microscopic situation of the network,and the middle layer can be used to explore the evolution path of the network. Compared with the traditional social network analysis,the proposed method can obtain more in-depth and detailed layer structure and internal correlation of knowledge network.