International collaboration in SSH: looking further than the Web of Science.

In 2019, a new parameter was added to the Flemish performance-based research funding system for universities to incorporate international collaboration in Flemish research publications. Therefore, the Flemish Academic Bibliographic Database for
the Social Sciences and Humanities (VABB-SHW), compiling academic SSH publications and serving as an instrument for funding distribution, required an extension with author affiliation data for all co- authored SSH publications. As less than half of the targeted VABB-SHW publications are included in the Web of Science (WoS), author affiliations had to be collected alternatively for the remaining publications (almost 25,000 for a ten-year time window). As a result, the implementation of the internationalization parameter not only led to data about Flemish international SSH collaboration outside the WoS, but also delivered a methodology for a multifaceted affiliation data collection operation.

The arrival of complementary author affiliation data enhances insight in international co- authorship in the Flemish academic SSH landscape. The nature of VABB-SHW – a strict differentiation between WoS and non-WoS publications - allows a mutual comparison on multiple levels, among them academic discipline, language and publication type. As the amount of international collaboration in SSH is much debated, author affiliation data for a broader range of publications, including those beyond the WoS, can now provide a more precise idea of the degree of international collaboration in terms of co-authorship.

The project showed that the retrieval of author affiliation data for non-WoS publications requires a multidimensional approach. Matching with alternative citation databases (Scopus, Crossref, …) is one component of the strategy, but by far the only
one. Web searches, consulting university repositories, visiting academic libraries and even direct communication with authorsare necessary steps to obtain a comprehensive dataset. Furthermore, data retrieval goes hand in hand with consistent data registration: all affiliated organizations were mapped to unique (ROR) organization identifiers. Missing links were discovered as a substantial part of the organizations could not be coded with the available set of identifiers. A codification of the methodology, both for retrieval and registration, may contribute to assist future projects.