Impact of a reclassification of Web of Science articles on bibliometric indicators

TL;DR

This study evaluates a new article reclassification method for Web of Science that enhances classification accuracy and significantly impacts bibliometric indicators, affecting disciplinary profiles and journal choice analyses.

Contribution

It introduces a reclassification approach based on paper references that improves classification accuracy and alters bibliometric indicators at the article level.

Findings

Reclassification improves classification accuracy and modularity.

Bibliometric indicators are significantly affected by the new classification.

Country-specific journal selection patterns are identified.

Abstract

This work aims at evaluating a reclassification of Web of Science articles implemented at OST. Articles from the 254 scientific categories of the Web of Science were reclassified at article level in 242 modified categories and 11 disciplines using the method of S. Milojevi\'c (2020). The reclassification is based on paper references categories and it no longer assigns papers to multiple or to multidisciplinary categories. It improves the accuracy and the modularity of the WoS classification. As there are important changes in document assignment at the lowest level, usual indicators such as disciplinary profiles or field normalized indicators are significantly modified. This study examines some of these modifications to provide explanations for the recipients of OST reports. Changes in specialization indexes reveal specific journal choices by scientists. In a sample of 25 countries, Brazil and China offer examples of facilities or constraints for selecting journals to publish certain research works.