PERKS project aims to enable more efficient, effective procedural knowledge management and procedure execution through semantic modeling of procedural knowledge and AI applications that make use of this knowledge. To achieve this, a key part is the PERKS architecture which is the key to ensure the proper operation of the three pilot use cases (BEKO Europe, Fagor Automation, and Siemens) At this time, an initial design of the PERKS solution architecture has been developed and in the following months, it will be tested in the use cases.
PERKS architecture will address requirements in seven main categories: Procedure Modelling, Procedural Knowledge Extraction, Procedure Validation and Approval, Procedure Execution, Procedure Storage Management, Procedural Knowledge Auditing, Procedural Knowledge Interface.
The technical team in charge of the design of PERKS architecture has been working on it in the first months of the project to start to apply it in the use cases.
We have had the pleasure of talking with this technical them to learn more details:
What does the architecture consist of?
The PERKS project design has adopted a Service-Oriented Architecture (SOA) to ensure independent, modular components capable of managing complex functionalities and scalability. This approach allows components to interact through APIs and other message-exchange mechanisms, promoting modularity, interoperability, and scalability within the system.
Several key decisions were fundamental to shaping the PERKS system’s architecture and are the common denominator of the proposed solutions:
- Using Semantic Web technologies and Knowledge Graphs (KGs) for effective knowledge management
- Applying data management design patterns to ensure data consistency and flexibility
- Utilizing a Service-Oriented Architecture (SOA), which decouples the system’s business capabilities
- Leveraging Large Language Models (LLMs) and state-of-the-art AI technologies to support natural language interactions and knowledge retrieval
How has the design and development process been?
In defining the PERKS architecture, we used the Arc42 methodology, which provided a structured template to document the system’s architecture. Following this, our documentation included several main elements: the general context, solution strategy, building blocks (organized into two levels – a black box level view and a white box level view), runtime views, technical and cross-cutting concerns, and deployment views. This was a collaborative effort and included both technical partners and use case representatives, ensuring the architecture addressed specific needs across all industrial contexts.
We began by establishing a comprehensive set of architecture requirements that guided the PERKS solutions’ structure and behavior.
- The initial step was creating a black box view of the PERKS system, a high-level overview showing individual solutions covering different aspects.
- After establishing this high-level structure, we developed a white box view for each of the six solutions, highlighting internal components and dependencies.
- Beyond structural definitions, we mapped out the system’s behavior by detailing the interactions within each solution
- We defined Technical and Cross-cutting Concerns which capture topics spanning multiple solutions, including:
- Defining the ontology and conceptual model for knowledge representation,
- Establishing the lifecycle of procedures within the system,
- Communication strategies between components for optimal performance.
- Considering specific technological requirements of the solutions forming the PERKS system, we have defined an hybrid deployment approach to combine solution deployment on premises and integration with commercial solutions on the Cloud.
- Additionally, we adopted best practices for development and release management, utilizing GitHub and CI/CD tools to streamline the development lifecycle and ensure consistent release quality.
How will this architecture improve the PK? What capabilities does the architecture offer?
Some key capabilities offered by PERKS architecture are:
- Reducing the time and effort required to document and maintain procedures
- Minimizing errors in procedure execution through better guidance and validation
- Preserving expert knowledge in a structured format
- Enabling more efficient training and knowledge transfer
- Supporting continuous improvement through better visibility into procedure execution and outcomes
The PERKS Service-Oriented Architecture (SOA) enhances procedural knowledge management by breaking down the system into independent, flexible services. Each service – whether it’s for knowledge extraction, validation, or user interaction – can be customized, updated, or integrated with existing company systems without affecting the others. The independence of services also means better maintenance and updates, as issues in one area don’t impact the entire system’s operation.
Did you encounter any challenges during development or implementation?
One of the main challenges we encountered was the diversity across industrial use cases—each originating from completely different domains, including manufacturing, energy management, and machine automation. This variance required the PERKS system to be flexible enough to support a wide range of procedures, data types, and operational needs.
While challenging, this diversity also presented a unique advantage. By addressing the distinct requirements of each domain, we designed a system adaptable to various industrial contexts.
Listen to our insightful podcast here to learn more about the PERKS architecture.
You’ll be able to explore our deliverable ‘D4.1 Reference architecture for a Procedural Knowledge Management System’ very soon. In the meantime, visit our Zenodo community to discover everything about PERKS!