As the Indonesian government advances digital document management through the SRIKANDI system, challenges persist regarding fragmented and subjective classification practices. This study proposes the integration of Natural Language Processing (NLP)-based classification within SRIKANDI to enhance consistency, transparency, and accountability in document management. Framed by an interdisciplinary theoretical foundation, the study synthesizes Michael Buckland’s document theory, viewing documents as dynamic social evidence, with Michel Foucault’s theory of power, highlighting classification as an exercise of institutional authority, and NLP methodologies that enable automated, content-driven categorization. The study positions documents as both technological artifacts and political constructs, whose classification practices simultaneously structure meaning and reinforce power relations. To assess the feasibility and implications of this proposed integration, the research adopts a qualitative approach, employing a case study of the SRIKANDI platform complemented by in-depth collection and analysis of practitioner feedback. Critical analysis reveals that classification systems risk reproducing institutional hierarchies if epistemological and political dimensions are not explicitly addressed. This study invites practitioners to evaluate the proposed NLP-enhanced system, offering reflections on its technical utility, potential biases, and impact on bureaucratic transparency. The insights gathered aim to inform the future development of more equitable, accountable, and semantically robust governmental information systems.
Sofiyani et al. (Mon,) studied this question.