Fig. 2

Example illustration of the data collection process. The figure depicts the typical minimum steps which are necessary to produce a dataset for TDM of scientific literature. Depending on the use case, tens or hundreds of different data sources may need to be accessed, each potentially requiring a different process–for example accessing a different set of API methods or a different process for downloading publication full text. Furthermore, depending on the use case, additional steps may be needed, such as extraction of references, identification of duplicate items or detection of the publication’s language. In the context of CORE, we provide the details of this process in Section Methods.