期刊名称：Iranian Journal of Information Processing & Management
印刷版ISSN：2251-8223
电子版ISSN：2251-8231
出版年度：2013
卷号：28
期号：2
页码：411-432
出版社：Iranian Research Institute for Information and Technology
摘要：Attribute selection is one of the steps before patent clustering. Various attributes can be used for clustering. In this study, the effect of using citation and citation title words, respectively, in form of bibliographic coupling and citation title words sharing, were measured and compared with each other, as patent attributes. This study was done in an experimental method, on a collection of 717 US Patent cited in the patents belong to 977/774 subclass of US Patent Classification. Fuzzy C-means was used for patent clustering and extended BCubed precision and extended BCubed recall were used as evaluation measure. The results showed that the clustering produced by bibliographic coupling had better performance than clustering used citation title words and existence of cluster structure were in a wider range of exhaustivity than citation title words.
