Evaluating JATS XML editing: an analysis of errors in published journal articles

Abstract

This paper analyzes errors found in the JATS XML editing of scientific articles already published in academic journals. The errors are categorized into seven types, along with an additional discussion about the use of generative artificial intelligence for JATS production. Identified issues include JATS files that fail to validate against the DTD; lack of the expected structure and misuse of elements and attributes defined by the standard; incomplete JATS files with discrepancies compared to the same articles in other formats; and problems related to language declaration, writing quality, and the use of deprecated elements. The study also examines errors in the commonly used OJS plugin that renders JATS content in HTML format. Special attention is given to the implications and risks of relying on generative artificial intelligence for creating JATS files. The findings highlight the need for significant improvements in the JATS editing process. Given how the analyzed JATS files have been done and published, the paper raises questions about the effectiveness of the interoperability that JATS aims to provide, and about the internal and/or external resources used and spent for its production.