Title:Journal subject classification: intra- and inter-system discrepancies in Web Of Science and Scopus

Abstract:Journal classification into subject categories is an important aspect in scholarly research evaluation as well as in bibliometric analysis. Journal classification systems use a variety of (partially) overlapping and non-exhaustive subject categories which results in many journals being classified into more than a single subject category. As such, discrepancies are likely to be encountered within any given system and between different systems. In this study, we set to examine both types of discrepancies in the two most widely used indexing systems - Web Of Science and Scopus. We use known distance measures, as well as logical set theory to examine and compare the category schemes defined by these systems. Our results demonstrate significant discrepancies within each system where a higher number of classified categories correlates with increased range and variance of rankings within them, and where redundant categories are found. Our results also show significant discrepancies between the two system. Specifically, very few categories in one system are "similar" to categories in the second system, where "similarity" is measured by subset & interesting categories and minimally covering categories. Taken jointly, our findings suggest that both types of discrepancies are systematic and cannot be easily disregarded when relying on these subject classification systems.