From narrative to machine-readable logic: Formalising and validating local indigenous knowledge for rural early warning systems

Abstract

Technological-based early warning systems (EWS) in rural Indonesia have shown limited long-term adoption because centralised, top-down mechanisms fail to incorporate the contextual triggers trusted by local communities, resulting in a disconnect in how warnings are understood, leading to inappropriate responses. Field evidence reinforces this problem; the usage of Information and Communication Technology (ICT)-based disaster information systems remained below three-quarters of users and the perceived benefits were also limited. At the same time, Local Indigenous Knowledge (LIK) has long played a critical role in disaster preparedness in rural communities. Local Indigenous Knowledge demonstrated universal adoption and was consistently considered more useful in preventing loss of life, fishing gear, catch and boats. Although prior studies have attempted to integrate LIK into disaster technologies, existing frameworks rely heavily on expert-driven knowledge extraction from qualitative interviews. This approach produces expert systems with unvalidated rules, limiting their credibility and scalability. To address this gap, this study proposes a socio-technical integration framework that systematically incorporates LIK into EWS. The framework was developed through a qualitative inquiry involving 17 in-depth interviews with fishermen and a focus group discussion (FGD) with eight fishermen community leaders from three coastal provinces of Indonesia, and its contextual foundation was further validated using 438 fishermen survey across the same regions.
Contribution: The framework introduces three stages: (1) LIK acquisition from community experience, (2) validation through empirical community consensus and scientific explanation and (3) structured integration into EWS. This staged community validation approach reduces dependence on tacit expert judgement and supports future integration with data-supported decision processes.