Research on the Construction of Capability Requirement Indicators for Training Simulation System with GBDT-Driven Questionnaire Mining

As intelligent technologies reshape instructional paradigms in the defense and training domains, simulation-based systems have become essential for enhancing educational effectiveness and supporting complex scenario rehearsal. This study aims to clarify what such systems should deliver by identifying and validating key capability requirements from both institutional and user-centered perspectives. A capability-based analysis approach is adopted to construct a five-domain indicator framework covering environment simulation, role modeling, procedural execution, performance evaluation, and operational support. A structured questionnaire was designed to collect user perceptions, and a Gradient Boosting Decision Tree (GBDT) model was applied to extract feature-level insights. Despite a modest R<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> of 0.2001, the model highlighted safety, instructional effectiveness, and innovation as top user concerns. The findings affirm that theoretical design goals align with user priorities and show that machine learning can help validate system needs and guide future development in data-driven defense training.