【摘要】 为解决科研重复立项及低水平重复研究问题,分析了科研项目文本的特点,提出了一种改进的模糊聚类算法RM-FCM。该算法充分考虑不同属性的特征项对科研项目分析的支持度,先用科研领域作为初始聚类中心,用参数调节科研文本各部分的重要性,然后不断计算各部分对中心的隶属度来确定聚类中心,最后通过聚类分析挖掘项目的相似性和项目间蕴涵的关联关系。实验结果表明,RM-FCM算法不仅准确地找出相似项目,也可以找出创新项目和交叉学科项目。更多还原

【Abstract】 In order to resolve the problems about the duplication of research projects and low-level repeated research in research mana-gement,by analyzing the characteristics of the text of the research projects,an improved fuzzy clustering algorithm RM-FCM is presented.The characteristics of the different attributes which affect the accuracy of the analysis for research projects is fully consided.Firstly,taken the field of scientific research as the initial cluster centers,parameter is used to adjust the importance of the various parts of the research text.Secondly,the various parts of the centers’ membership is calculated to determine the cluster center.Finally,the similarity and the implied association among projects is researched by cluster analysis.Experimental results show that RM-FCM algorithm can effec-tively and accurately find similar projects,innovative projects and interdisciplinary projects.更多还原