RetractBASE (retractbase.csic.es), the largest open search engine specialized on retracted literature, makes publicly accessible their datasets here: osf.io/xtrsb/files/...

Today, I presented retractbase.csic.es, our open database about retractions, in Bristol STI-ENID Conference. @stienid2025.bsky.social with a very good reception

RetractBASE also incorporates author, organization and journal indexes to identify the entities with the most retractions and enrich the data exploration.

It includes more than 120k publications between 2000 and 2024, where all they are connected with their retraction notices, and external links to the original publication, other databases and @pubpeer.com are provided to broadering information about the reason of retraction.

5/7 Regarding document types, all the databases have problems in the extraction and identification of entities in books and book chapters. Books (.31) show greater author variability, and improper extraction is seen in affiliations (15.9%) and venues (44.7%)

3/7 Crossref-based products such as Dimensions, Scilit, and The Lens identify more venues and publishers than others, while they have limitations in the identification of authors and orgs. Dimensions is the database with the best performance, in comparison to Scilit and The Lens.

2/7 Dimensions and OpenAlex are the best products processing authors because they have the lowest percentage of authors with one publication (>88%), and the lowest slope coefficient. They also show low average author variation (Dimensions, .12; OpenAlex, .17).

1/7 “Research entity information and coverage in eight free access scholarly databases” is our (Lorena Delgado and me) last comparison of free-access scholarly databases, according to four research entities (authors, organizations, venues and disciplines). osf.io/preprints/so...

9/9 However, and despite pseudoscience communities could have a marginal position in mainstream science; it’s worrying to notice that they have better social impact than scientific results, suggesting that it could be more acceptable to public opinion, being a fundamental pillar of its prevalence.

8/9 Regarding the correction, there are no differences when editorial notices are released, but the number of reported papers on PubPeer is considerably higher in pseudoscientific disciplines, indicating that their methodologies are not acceptable for a scientific standard and then are reported.

7/9 Linked to this, there is a great presence of professional associations and non-profit organizations performing and funding pseudoscience studies. While in scientific fields predominate universities and government agencies, excepting integrative medicine.

6/9 The most remarkable difference is the important presence of practitioners and self-employed researchers in pseudoscientific studies (authors and editorial board's members), suggesting that these authors lack of experience in research activities and perform studies to validate their practices.

4/9 the percentage of journal-self citations in pseudoscience are higher than in science papers, while the proportion of cross-group citations show an inverse trend. This finding suggest that pseudosciences tend to be more isolated and more resistant to external ideas than scientific communities.

1/9 “How does pseudoscience differ from science? A pair-wise bibliometric analysis” is my first approximation to the study of pseudoscience from a bibliometric approach, observing differences between pseudoscientific and scientific disciplines using publishing parameters doi.org/10.31235/osf...

6/6 References from publications by Georg Thieme, Routledge and, to a lesser extent, Oxford University Press, have trouble being indexed in all databases. This fact emphasizes the importance of book publishers providing references and ensuring their indexation in databases.

5/6 the study of references has detected how specific document types, such as books and book chapters, present limitations when their references are extracted and indexed in scholarly databases. This limitation is transversal to every database, suggesting that these issues are caused by publishers.

4/6 Analysis of references has revealed that Scilit only extracts references with Digital Object Identifiers (DOI) and that Semantic Scholar causes significant problems when it adds references from external web versions. Microsoft Academic was updated several years previous to the defunct.

3/6 Publications in Crossref-based databases (Crossref, Dimensions, Scilit and The Lens) have similar citation counts, while in search engines (Google Scholar, Microsoft Academic and Semantic Scholar) they have more citations due to a greater coverage of publications and integration of web copies.