Generalization of bibliographic coupling and co-citation using the node split network

Bibliographic coupling (BC) and co-citation (CC) are the two most common citation-based measures of similarity between scientific items. One can interpret these as second-neighbor relations distinguished by direction citation: BC is a citing items, whereas CC that cited A previous study proposed two-layer node split network emulate clusters in computationally efficient manner; however, lack intralayer links makes it impossible to obtain exact similarities. Here, we propose novel methods estimate on using personalized PageRank (PPR) neural embedding (EMB). We demonstrate PPR strongly correlated with measures. Moreover, our method yield precise similarities items even if they distant from each other. also show many high missing original BC/CC network, which suggests essential consider long-range Comparative experiments global local edge sampling suggest stable for networks. This analysis offers valuable insights into process searching significantly related regarding measure.