TL;DR
Research groups often create powerful tools that remain proofs-of-concept due to a lack of industrial-grade polish. This paper argues that the most effective strategy for turning research prototypes into widely-used, high-quality tools is to partner with a technology provider for industrialization, as demonstrated by the ATL model transformation language.
What Problem Does This Paper Address?
Many academic research groups develop sophisticated tools to address real-world industrial problems. While these tools are often technically sound and innovative, they rarely achieve widespread adoption by industry practitioners. This gap exists because research prototypes typically lack essential "commercial-quality" features such as robust user interfaces, comprehensive documentation, reliable user support, and long-term maintenance. Research teams, constrained by limited resources and funding priorities that favor novel research over tool refinement, struggle to bridge this gap. This paper addresses how research groups can overcome these limitations to make their tools impactful and widely adopted.
Methods
The paper proposes a two-pronged strategy for industrializing research tools:
- Technical Design:
- Modularity: The ATL framework is built on a modular structure leveraging Eclipse's OSGi platform. This allows for clear separation of concerns, making it easier for independent actors to contribute to specific components (e.g., compiler, virtual machine) without needing to understand the entire system.
- Standard Technologies: ATL adopts widely-used, open-source technologies like Eclipse and the Eclipse Modeling Framework (EMF). This leverages existing infrastructure, facilitates integration, and benefits from a large community.
- Interoperability: The architecture is designed for independence from specific modeling layers, allowing ATL to be ported to different environments and ensuring its longevity beyond any single technology.
- "Eating Your Own Dog Food": The ATL development team uses their own tool extensively, which helps in testing and improving it organically.
- Business Model:
- Partnership with a Technology Provider: The core of the strategy is forming a partnership with a technology provider (like Obeo for ATL). This provider takes responsibility for non-core research aspects, such as:
- Quality assurance and bug fixing.
- Ensuring interoperability with other tools.
- Maintaining continuity and backward compatibility.
- Improving user experience and interface.
- Managing release cycles.
- Providing user support (free and commercial).
- Dual-Branch Development: A "Research" branch of the tool is maintained for experimental features, while an "Official" branch, managed by the technology provider, focuses on stability, quality, and industrial readiness. This allows for rapid research iteration without compromising the stable version.
- Open Source Licensing: Using an open-source license (like the Eclipse Public License) facilitates collaboration and technology transfer between the research lab and the provider, and encourages contributions from the wider community.
Datasets
Not discussed in the paper. The paper focuses on the development and industrialization process of a specific tool (ATL) rather than empirical evaluation on datasets.
Key Insights and Ideas
- The "Commercial-Quality" Gap is Real and Costly: Research tools often fail to gain traction not due to technical flaws, but due to a lack of polish and support that industry demands. This represents a missed opportunity for research impact.
- Partnership is Key to Industrialization: Research groups alone cannot typically provide the resources needed for industrialization. A dedicated technology provider is essential for handling non-core but critical aspects like support, documentation, and usability.
- Dual-Branch Development Balances Innovation and Stability: Maintaining separate "Research" and "Official" branches allows for cutting-edge experimentation while ensuring a stable, reliable product for industrial users.
- Open Source as a Facilitator, Not a Solution: While open source is crucial for collaboration and transparency, it doesn't automatically solve the industrialization problem. A deliberate business model is still required.
- A "Triangle" Model for Collaboration: The paper proposes a model involving a Research Lab, a Technology Provider, and Big Companies/User Communities to align interests and facilitate the industrialization process.
Key Results
The paper doesn't present quantitative results in terms of performance metrics or adoption rates. Instead, it highlights the success of this strategy through the evolution of the ATL tool:
- ATL's Growth and Adoption: The paper describes how ATL, initially a research prototype, has evolved into a full Eclipse project with a significant user community.
- Successful Partnership: The collaboration with Obeo has allowed ATL to achieve industrial quality, attract company users, and provide a stable base for future research.
- Generalizability: The authors are replicating this industrialization strategy with other tools from their group (MoDisco, Megamodeling), indicating confidence in its effectiveness.
Limitations and Future Work
- Formalizing Partnerships: The authors acknowledge the need to formalize the protocols for establishing and executing these partnerships to clarify roles, resources, and responsibilities.
- Collaboration Platform: They plan to create a web-based collaboration platform for ATL extensions ("ATL Labs") to facilitate experimentation and contributions beyond the core partner.
- Community Building: The authors intend to leverage social networking tools to build a stronger ATL user community, as existing Eclipse mechanisms are deemed insufficient.
- Generalizability: While the strategy is presented as generalizable, the paper focuses on the ATL case study. Further validation across different tool types and domains would strengthen the argument.
Summary updated