In this paper, we study how visual workflows are reported in scientific publications and how incomplete workflow preservation can make older studies harder to reproduce. Using KNIME as a case study, we develop and apply a semi-automated evidence pipeline that combines OpenAlex bibliometrics, a 100-record top-cited article audit registry, workflow retrieval, and opening and execution checks in a current KNIME environment. The OpenAlex search shows that KNIME is visible across a broad scientific literature, while the article assessment shows that workflow figures and textual descriptions are more common than downloadable workflow artifacts. Of the 100 audited records, 22 report downloadable or linked KNIME workflows, and recorded retrieval attempts obtained workflow artifacts or workflow directories for 18. In the recorded manual checks, five article records had at least one workflow execute successfully. These results show that incomplete artifact and environment preservation limits the long-term reproducibility of KNIME-based studies.
Vitalii Kaplan (Tue,) studied this question.