The main objective of this study is to evaluate the coverage and information quality of research entities (authors, organizations, venues and disciplines) in eight new academic databases.
A random Crossref sample of over 115k document object identifiers was chosen and subsequently searched across seven databases.
Dimensions and OpenAlex are the best products processing authors because they have the lowest percentage of authors with one publication (Dimensions, 88.1%; OpenAlex, 89.9%) and the lowest slope coefficient (old OpenAlex, a = 3.25; Dimensions, a = 3.46). They also show low average author variation (Dimensions, 0.12; OpenAlex, 0.17). Microsoft Academic is the database that detects the most affiliations (87%) and organizations (71.2%). Crossref-based products such as Dimensions (98.1%), Scilit (99.3%) and The Lens (96.4%) identify more venues and publishers than other products. Semantic Scholar is highlighted as the database that thematically classifies the most publications (94.1%). Regarding document types, the study also identifies transversal problems in the extraction and identification of entities in books and book chapters.
The results of this study have important implications for selecting different databases when it comes to searching literature for reviews, meta-analyses and other studies.
This is the first study that compares the largest number of free-access scholarly databases, exploring the completeness degree and quality of the research entity information (authors, organizations, disciplines and venues).
