论文部分内容阅读
[目的 /意义]基于高维矩阵稀疏降维的思想,提出一种利用惩罚性矩阵分解(Penalized Matrix Decomposition,PMD)实现共词分析的新方法。[方法 /过程]以“学科服务”为研究主题,根据PMD算法原理,在Matlab环境下分别实现特征词的提取、特征词的软聚类以及聚类效果的可视化。[结果 /结论]与传统的共词分析方法对比,PMD算法在共词分析中具有独特的优势:提取的特征词比较全面,聚类数目便于确定,聚类结果易于理解。
[Purpose / Significance] Based on the idea of sparse dimensionality reduction in high-dimensional matrices, a new method of co-word analysis using Penalized Matrix Decomposition (PMD) is proposed. [Methods / Processes] With the subject of “Subject Services” as the research topic, according to the principle of PMD algorithm, the feature words are extracted, the soft clustering of feature words and the visualization of clustering effects are respectively implemented under the Matlab environment. [Result / Conclusion] Compared with the traditional co-word analysis, the PMD algorithm has unique advantages in co-word analysis: the extracted feature words are more comprehensive, the number of clusters is easy to be determined and the clustering results are easy to understand.