| 作者共现网络的科学研究结构揭示能力比较研究 |
| |
| 引用本文： | 邱均平,董克.作者共现网络的科学研究结构揭示能力比较研究[J].中国图书馆学报,2014,40(1):15-24. |
| |
| 作者姓名： | 邱均平 董克 |
| |
| 作者单位： | 武汉大学信息管理学院，湖北 武汉 430072;武汉大学信息管理学院，湖北 武汉 430072 |
| |
| 基金项目： | 本文系国家社科基金重大项目“基于语义的馆藏资源深度聚合与可视化展示研究”(编号:11&ZD152)的研究成果之一。 |
| |
| 摘 要： | 作者共现网络在科学研究领域结构分析中具有非常重要的作用，本文构建了作者合作、作者共被引、作者文献耦合、作者关键词耦合及作者期刊耦合五种网络，利用层次聚类和网络关联分析比较不同共现网络揭示科学研究结构的能力。通过层次聚类发现：作者文献耦合网络的准确性和精确性最高，作者共被引网络和作者合作网络其次，作者关键词耦合网络的结果容易受到干扰，作者期刊耦合网络的分析效果较差。通过网络关联分析发现：作者合作网络与其他网络的相关性都相对较低，与其他网络发现的角度不同；作者文献耦合网络与其他几种网络之间的相关性最高，对其他网络分析结果的替代性较强，多种方法结合运用更能反映科学研究的实际结构。本文还讨论了共现网络子结构研究中凝聚子群分析及中心性指标的适用性问题。图5。表4。参考文献16。
|
| 关 键 词： | 作者共现网络 科学结构 子结构分析 关联分析 数据清洗 |
| 收稿时间： | 2013/3/27 0:00:00 |
| 修稿时间： | 2013/4/18 0:00:00 |
A Comparative Study on the Ability of Author Co-occurrence Network in Revealing Scientific Structure |
| |
| Qiu Junping and Dong Ke.A Comparative Study on the Ability of Author Co-occurrence Network in Revealing Scientific Structure[J].Journal of Library Science In China,2014,40(1):15-24. |
| |
| Authors: | Qiu Junping and Dong Ke |
| |
| Abstract: | Co-occurrence network at author-level plays a pivotal role in the structural analysis of intellectual and scientific domains.Five kinds of co-occurrence networks at author-level were constructed in this paper:co-authorship, author co-citation, author bibliographic coupling, keyword-based author coupling, journal-based author coupling.Comparative analysis of hierarchical clustering results showed that author bibliographic coupling, followed by author co-citation, co-authorship, keyword-based author coupling and journal-based author coupling, was proved to be the most accurate and effective way for discovering sub-structure of scientific research.QAP correlation analysis showed that the correlation of co-authorship with other networks was the lowest, and had a different occurrence perspective from other networks, while the correlation of author bibliographic coupling with other networks was the highest and had stronger substitutability for results of other networks.The multipe use of networks is suggested as a better choice for finding real structure of scientific research domains.Finally the paper discusses some specific problems about the fitness of cohesive subgroups analysis and centrality indicators for sub-network analysis in author co-occurrence networks.5figs.4tabs.16refs. |
| |
| Keywords: | Author co-occurrence network Scientific structure Sub-network analysis Correlation analysis Data cleanup |
|
| 点击此处可从《中国图书馆学报》浏览原始摘要信息 |
| 点击此处可从《中国图书馆学报》下载免费的PDF全文 |
