Accuracy of affiliation information in Microsoft Academic: Implications for institutional level research evaluation

Authors: Ranjbar-Sahraei B.; Eck, N.J. van; Jong R. de

Comment: This is a summary of results for a poster presented at the STI 2018 Conference in Leiden. The work compares research output recorded by both Microsoft Academic (MA) and Web of Science (WoS) for Leiden University. A first level automated matching is done, revealing differences across MA and WoS. Then, a sample of 100 is drawn from each of the disagreeing parts of the comparison. Manual checking of these found that MA contained affiliation errors.

Abstract: In this work, we study the accuracy of affiliation information in Microsoft Academic (MA). To conduct this study, we have considered the full set of publications assigned to Leiden University (LU) as provided by two different data sources: MA and Web of Science (WoS). The results of this study suggest that a considerable number of publications in MA have missing or wrong affiliation information.

Source: Accuracy of affiliation information in Microsoft Academic: Implications for institutional level research evaluation