摘要
为了发现企业技术实力和行业趋势,减少不必要的研发成本、做出正确决策,给出了基本专利同被引矩阵分析方法。利用改进的粗糙集K-Means模糊聚类方法实现对不同主题专利聚类,解决了重复计算中心向量带来的不准确性;进一步利用关联规则挖掘算法发现强关联规则,以强关联规则结论为该类别核心专利,提高针对性地选择专利。实例表明了该方法的有效性,为专利引文分析提供了可行的解决途径。
To find the enterprise technical strength and industry trends, decrease the cost for researching and make the right decision, an analytical method based on basic patent co-citation matrix is proposed. This method used improval k-means fuzzy clustering algorithm based on rough sets, realize different thematic patents cluster, and the problem of inaccuracy caused by double counting is resolved. Then, using association rule mining algorithm to discover strong association rules whose conclusions are their core patents, which can raise the pertinence for chosing the patents. Finally, the feasibility of the method is validated by practical application.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第8期1779-1781,1785,共4页
Computer Engineering and Design
关键词
专利文献
同被引
粗糙集
模糊聚类
关联规则
patent literature
co-citation
rough set
fuzzy clustering
association rules
