# Fault tree analysis-adapted knowledge structuring: a case study of sustainable international security cooperation

## TL;DR

This paper proposes a method using knowledge graphs and Fault Tree Analysis to improve sustainable knowledge transfer in Japan's international security cooperation.

## Contribution

A novel knowledge engineering method combining CHARM and Fault Tree Analysis for externalizing tacit expertise in international security cooperation.

## Key findings

- The method increased the volume and granularity of externalized knowledge, yielding 42 reference cases and 133 case-attached actions.
- FTA-adapted CHARM approach supports continuity during constrained handovers and improves negotiation preparedness.
- The approach generates reusable, machine-interpretable assets for human-AI collaboration.

## Abstract

Sustainable knowledge transfer in Japan’s international security cooperation for research and development (R&D) and procurement is challenging due to institutional and security constraints. Critical know-how is often tacit and dispersed among experts; continuity is undermined by frequent personnel rotations, temporal–spatial gaps between projects, and institutional and cultural differences across international partners, leading to knowledge loss. In this context, traditional on-the-job training is difficult to sustain, making durable knowledge transfer difficult to achieve. To overcome this problem, this study proposes a knowledge engineering method that externalizes practitioner expertise. Procedure-based knowledge (what/how) and purpose-based knowledge (why: purposes and decision rules) are structured as two auditable linked, machine-interpretable graphs—the procedure- and purpose-based knowledge graphs. This method uses the Convincing Human Action Rationalized Model (CHARM) as a notation for knowledge structuring. To articulate implicit causal reasoning, we integrate Fault Tree Analysis (FTA) as a qualitative, deductive elicitation and articulation notation. Starting from observed outcomes (“top events”), FTA deductively elicits avoidance purposes and candidate actions. These purpose-action pairs are then recorded under Reference Cases (RC) and embedded as RC-tagged links in the two graphs with FTA-derived annotations, thereby refining causal logic and facilitating knowledge externalization. We empirically assess the method’s effectiveness through qualitative and quantitative analyses of data from semi-structured interviews, a facilitated workshop, and FTA-guided follow-up interviews. These activities increased both the volume and granularity of externalized knowledge, yielding 42 RCs and 133 case-attached actions as provenance-bearing purpose-action units. Our approach yields reusable, machine-interpretable assets for human-AI collaboration and may support continuity during constrained handovers, which may help mitigate repeated errors and improve negotiation preparedness. These findings suggest that the FTA-adapted CHARM approach can foster more sustainable knowledge transfer for Japan’s international security cooperation.