Qualitative Coding in the Age of AI:An Ontology-Driven Approach

Qualitative coding is an essential methodological tool in qualitative research. Although various tools exist to support manual qualitative coding, the process remains highly resource-intensive, requiring significant time and expertise. It is further complicated by inconsistent reporting of coding protocols, differing interpretations of inter-coder reliability metrics, and difficulties in achieving conceptual agreement among coders – all of which contribute to the broader replication crisis in science. Meanwhile, large language models have demonstrated remarkable abilities to understand context across diverse tasks and are increasingly applied to information and, in combination with semantic web technologies, knowledge extraction. In this work, we define qualitative coding process as a knowledge base construction task and propose an ontology-mediated approach to automating qualitative coding, formalising key aspects of the methodology and reliability assessment using semantic web<br/>standards. We evaluate the reliability of this approach through a concrete implementation and case study, finding that qualitative coding can benefit substantially from AI-driven automation – especially when the underlying<br/>coding ontology is well-defined and domain-relevant constraints are explicitly codified.