Optimizing SCImago Journal & Country Rank classification by community detection

Clustering analysis is one of the most successful classification techniques used in bibliometrics and scientometrics, searching to develop reliable and consistent tools and outputs.

Abstract

Subject classification arises as an important topic for bibliometrics and scientometrics as to develop reliable and consistent tools and outputs. For this matter, a well delimited underlying subject classification scheme reflecting science fields becomes essential. Within the broad ensemble of classification techniques clustering analysis is one of the most successful.
\n
\nTwo clustering algorithms based on modularity, namely, VOS and Louvain methods, are presented in order to update and optimise journal classification of SCImago Journal & Country Rank (SJR) platform. We used network analysis and visualization software Pajek to run both algorithms on a network of more than 18,000 SJR journals combining three citation-based measures, that is, direct citation, co-citation and bibliographic coupling. The set of clusters obtained was termed through category labels assigned to SJR journals and significant words from journal titles.
\n
\nDespite of both algorithms exhibiting slight performance differences, the results showed a similar behaviour in grouping journals and, consequently, they seem to be appropriate solutions for classification purposes. The two new generated algorithm-based classifications were compared to other bibliometric classification systems such as the original SJR one and WoS Subject Categories in order to validate their consistency, adequacy and accuracy. Although there are notable differences among the four classification systems analysed, we found a certain coherence and homogeneity among them.