PlanXFlow
Overview
PlanXFlow is a software development lifecycle designed for engineering AI planning systems. PlanXFlow provides a structured, waterfall-like yet iterative approach tailored to the unique challenges in AI planning.
The lifecycle encompasses ten phases, guiding researchers and practitioners from conceptualization and design to deployment, monitoring, and analysis. It acts as a foundational guide that helps establish a common language among stakeholders and improves traceability and reproducibility in AI planning projects.
The Ten Phases
- Requirements Analysis: Identify functional, non-functional, domain, and user requirements for the planning system.
- AI Planning Model Formulation: Define a suitable planning model based on planning type, world context, and user features.
- Domain Model Design: Formalize domain knowledge using planning languages such as PDDL, HPDL, or HDDL.
- System Design and Architecture: Design architecture and workflows that integrate planning functionalities and services.
- Planning Technology Selection: Evaluate and select existing planning tools before building new components.
- Implementation: Develop planning components, communication mechanisms, and supporting artifacts.
- Testing: Validate requirements with both isolated and integration testing across domains and workflows.
- Deployment: Install and configure components for reliable execution environments.
- Monitoring: Collect and process provenance data to improve quality, reproducibility, and explainability.
- Analysis: Analyze provenance and system behavior to drive iterative improvement.
Innovative Elements
- Planning Model: Captures the interplay among planning type, world context, and user features.
- Technology Selection: Emphasizes systematic selection of planning technologies.
- Provenance Data: Supports traceability of data sources and transformations throughout the lifecycle.
Core Features
- Adaptability: Tailor phases to specific project needs.
- Scalability: Expand or condense lifecycle phases by project scale.
- Transparency: Maintain clear and understandable process steps.
- Future-Proof: Keep workflows easy to upgrade over time.
Engagement and Contribution
Your experience and insights are valuable to the evolution of PlanXFlow. Whether used in industry or academia, feedback helps improve the lifecycle for the broader community.
For industrial applications, share how PlanXFlow was integrated and adapted in your systems. For academic use, please cite the following references.
- Georgievski, I. Software Development Lifecycle for Engineering AI Planning Systems. International Conference on Software Technologies, pages 751-760, 2023.
- Georgievski, I. Conceptualising Software Development Lifecycle for Engineering AI Planning Systems. IEEE/ACM International Conference on AI Engineering - Software Engineering for AI, pages 88-89, 2023.
BibTeX
@inproceedings{georgievski2023softtech,
title={Software Development Lifecycle for Engineering AI Planning Systems},
author={Georgievski, Ilche},
booktitle={International Conference on Software Technologies},
pages={751--760},
year={2023}
}
@inproceedings{georgievski2023caine,
title={Conceptualising Software Development Lifecycle for Engineering AI Planning Systems},
author={Georgievski, Ilche},
booktitle={IEEE/ACM International Conference on AI Engineering -- Software Engineering for AI},
pages={88--89},
year={2023}
}