Data sources

Bifrost® draws publication data from three complementary sources. Which source is used depends on the purpose of the analysis.

SwePub

SwePub is the National Library of Sweden's national compilation of Swedish research publishing. SwePub aggregates publication metadata from all Swedish DiVA repositories and from institutions that are not part of the DiVA consortium (for example the University of Gothenburg and Lund University), which makes it the best source for national overviews.

Bifrost uses SwePub for:

Limitations: The SwePub bibliometrics API lacks citation data, dissertations and certain personal identifiers that are available in the DiVA instances. SwePub only covers publications registered by Swedish institutions.

DiVA

DiVA is the shared publication database for a large number of Swedish institutions. Bifrost uses DiVA when an analysis requires information that SwePub does not provide, provided the institution in question is part of the DiVA consortium:

Limitations: DiVA lacks citation data and international coverage. Each institution administers its own DiVA instance with local metadata conventions, and institutions outside the consortium (e.g. GU and LU) are not present in DiVA.

OpenAlex

OpenAlex is an open database of research publications with broad international coverage. Bifrost uses OpenAlex for analyses that require citation data and cross-border comparisons:

Limitations: OpenAlex has broad but uneven coverage: the humanities and social sciences are systematically less well represented than the natural sciences and medicine, which affects the reliability of citation analyses for these fields. Subject classification follows OpenAlex's own system (based on the OECD Fields of Science), not UKÄ/HSV.

Enrichment

After the publications have been retrieved from SwePub or DiVA, Bifrost supplements each record with data from external sources. Enrichment makes it possible to analyse citation patterns, journal quality and open availability even for reports based on Swedish sources.

Per-source caching reduces the load on external services: NPI is cached for 30 days, SCImago for 365 days, OpenAlex per DOI.