Normalized Similarity Index: An adjusted index to prioritize article citations

One of the main applications of citation is to find articles that are relevant to a particular article. However, not all citations are equally relevant to the target article. This paper presents an approach to identify the most relevant citation(s). To this end, the Normalized Similarity Index (NSI) is proposed to quantify the similarity between the source and target of a citation base on the co-citations and references shared by them. To validate the method, NSI was calculated for five citation networks and was compared with the peer review grades for the relevancy between the source and the target articles. The results showed a significant correlation between the NSI ranks and those of peer review. Also, combined linkage (CL) and weighted direct citation (WDC) were calculated from the same data. According to the results of comparison between the NSI with other similarity measures, in most cases, NSI did better than others at reproducing the peer rankings. Our principal conclusion is that the NSI can be used to prioritize the citations of given highly cited article, and represent knowledge flow from the target article.

► Normalized Similarity Index can be used to prioritize the citations of highly cited article, and represent knowledge flow from the target article.
► We attempted to quantitatively measure the relevancy of citing and cited articles by using the direct and indirect citation linkages.
► NSI was calculated for five citation networks and compared to other similarity measures.
► The NSI can be generalized beyond the scopes of this study.