[院校资料]2011-11-18-基于分析工具的图书馆学科服务-张帆.ppt

* * * * * * * * * * This is how we define co-citation between a pair of papers. An individual citation is not an isolated event, but is embedded in a list of other cited references. We can use this fact to analyze the pattern of what is cited jointly, that is, how often cited works co-occur. When this co-occurrence is frequent, the association of ideas it represents reflects a group consensus. This slide is a schematic representation of how two papers become highly co-cited starting with one co-citation and moving to three. As an example, consider two well known nanoscience papers, one by Liu et al from Smalley’s lab at Rice University and one by Chen et al from Haddon’s group at the University of Kentucky. Each is cited over 300 times, but they are co-cited 160 times indicating a strong association. This slide also shows the formula we use to normalize the co-citation relationship. We can then use these patterns of association to perform a cluster analysis which is a process of aggregation of papers linked by co-citations. The cluster analysis gathers together the links that share common papers. For example, a link A-C is merged with a link C-E to form A-C-E and so on. Once the network of connected documents is formed, a map or visualization of each cluster can be created. Each co-citation link is modeled as an attractive force, and, in addition, there is a repulsive force among all objects. As an example, I have selected a large research front on carbon nanotubes from the 2003 period. Here we find the link between the Liu and Chen papers discussed previously. On the ESI web site we have an interview with Professor Richard Smalley of Rice University, the most cited author on this map. 自体骨髓干细胞;冠状动脉内骨髓源性祖细胞，人类心脏干细胞，成年心脏干细胞，心肌干细胞 * 共引论文 313 citations 374 citations Chen J, et al, Solution properties of single-walled Carbon nanotubes Science, 1998 Liu J, et al, Fullerene pipes Science,1998 160 co-citations 0.47 =160/(313*374)^.5 citation co-citation 聚类的构成 at