Mining Typical Features of Highly-Cited Papers

摘要

In this paper, the method to detect the future highly-cited papers (HCPs) in citation network was discussed. Considering the growing process of one paper, the content features describing the "rewards" that papers obtained in their earlier stage were extracted to characterize their quality mechanism. Integrating the content features and the external features obtained from the social view of papers' communication process, the feature space used to model HCPs was established. Basing on the feature space, the typical features of HCPs were extracted by the framework of rough set reduction. It shows that the papers' inner qualities and the external features mainly presented as the reputation of authors and journals make joint efforts to generating HCPs in future.