Modeling tool for managing canvas-based models traceability in ML system development

Abstract

Analysis of machine learning models often used canvas-based models such as ML Canvas and AI Project Canvas to facilitate rapid brainstorming of ideas. However, those models often cover only high-level descriptions of requirements. Developers may utilize other models to achieve a more comprehensive analysis to cover specific aspects. This condition may lead to inconsistencies between different models. This study proposes a tool to support traceability between canvas-based and other models. The tool is implemented as a plugin for astah* System Safety.