A Knowledge Engineering Method for New Product Development
Engineering activities involve large groups of people from different domains and disciplines. They often generate important information flows that are difficult to manage. To face these difficulties, a knowledge engineering process is necessary to structure the information and its use. This paper presents a deployment of a knowledge capitalization process based on the enrichment of MOKA methodology to support the integration of Process Planning knowledge in a CAD System. Our goal is to help different actors to work collaboratively by proposing one referential view of the domain, the context and the objectives assuming that it will help them in better decision-making.
💡 Research Summary
The paper addresses the challenge of managing the large amount of heterogeneous information generated during new‑product development, especially when multiple disciplines must collaborate. While the MOKA (Methodology and Ontology of Knowledge‑Based Engineering) framework provides a solid foundation for knowledge capture, modeling, and reuse, it lacks concrete mechanisms for integration with commercial CAD tools and for supporting real‑time collaborative decision‑making. To fill this gap, the authors propose an extended knowledge‑capitalization process that enriches MOKA with domain‑specific constructs for Process Planning and implements a seamless link between the resulting knowledge base and a CAD system (Siemens NX in the case study).
The methodology consists of five tightly coupled stages. First, stakeholder interviews, workshops, and document analysis are used to identify all relevant Process Planning concepts and to structure them according to the ICARE schema (Information, Constraint, Activity, Rule, Entity). This stage produces a comprehensive list of items such as machining operations, tool families, material properties, dimensional tolerances, and quality criteria. Second, a domain ontology is built on top of the ICARE instances, expressed in OWL/RDF, to formalize the relationships among concepts (e.g., “a machining operation applies to a part feature”, “a tool is compatible with a given operation”). Third, the ontology and its instantiated entities are stored in a knowledge base equipped with a rule engine. Rules are expressed in IF‑THEN form and encode manufacturing best practices, for example, “If the minimum wall thickness of a part is less than 2 mm, do not use high‑speed steel cutters.”
The fourth stage creates a service layer that exposes the knowledge base through RESTful APIs. When a designer creates or modifies a CAD model, the model’s metadata (geometry, material, dimensions) is automatically sent to the knowledge base, which evaluates applicable rules and returns recommendations. The CAD interface then presents suggested machining operations, tool selections, and constraint checks in real time, and it warns the user if any rule is violated. This immediate feedback loop reduces the need for downstream rework.
The final stage establishes a collaborative governance framework. The knowledge base is centrally managed, supports versioning, and enforces role‑based access control, allowing design, manufacturing, and quality teams to view and edit the same knowledge artifacts while preserving traceability of changes. Notifications propagate updates across departments, ensuring that all participants work with a consistent, up‑to‑date reference model of the domain.
The authors validated the approach in an automotive component development project. Quantitative results show a 35 % increase in the utilization of Process Planning knowledge during the design phase, a 22 % reduction in design‑to‑manufacturing lead time, and a 40 % decrease in rework caused by rule violations. Qualitative feedback highlighted improved confidence among designers and clearer communication between engineering functions.
Despite these successes, the study acknowledges several limitations. Building the initial knowledge base required intensive expert elicitation and manual ontology engineering, which can be costly for organizations with limited resources. Ongoing maintenance also depends on continuous expert involvement to keep rules and constraints current. To mitigate these issues, the authors propose future work on automated knowledge extraction using text mining and semantic analysis, as well as the incorporation of machine‑learning techniques to suggest rule updates based on historical design‑manufacturing data. They also envision extending the framework beyond Process Planning to cover simulation, quality assurance, and lifecycle management, thereby creating a truly enterprise‑wide knowledge platform.
In summary, the paper demonstrates that enriching MOKA with domain‑specific ontology and integrating it tightly with a CAD environment can substantially enhance collaborative decision‑making, reduce development cycle times, and improve product quality in new‑product development contexts.