An Example of Ontology-based Knowledge Sharing
in Concurrent Engineering

Junichi Yoshinaga, Machiko Chikano

Research & Development Headquarters, Yamatake Corporation

1-12-2 Kawana, Fujisawa, Kanagawa-Pref. 251-8522, JAPAN



A computer aided system, based on a framework utilizing systematized knowledge, is an effective way to vitalize intellectual activities of experts.

We proposed, and evaluated a conceptual design of a process design support system based on systematized knowledge in 1998. We have acknowledged that a system based on an ontology-and-multi-agent framework is effective to knowledge sharing and reuse. However, the more designers get involved, the more difficulties emerge by keeping and maintaining the logical consistency of ontology in the above framework.

To avoid such difficulties, we have reviewed a part of our conceptual design, regarding ontology structures and ontology-sharing interfaces.

Defining the functions to manage ontology-sharing interfaces is a key importance to get higher degree of efficiency in collaborative activities.

1. Introduction

Recently, manufacturers have confronted the need a new manufacturing paradigm that supports the entire engineering lifecycle and enables developments. This paradigm would be called the post-mass production paradigm, since it simultaneously contributes toward profitability and minimizes environmental impact.

The post mass production paradigm is possible by systems for manufacturing that have re-configurable features which enable closer alignment of the manufacturing practice with changes in the market and in environmental consciousness. Such manufacturing systems must be developed, designed, operated, and re-configured based on systematized and reusable manufacturing knowledge.

To realize a manufacturing system requires a process design. This is a series of design activities that visualize target costs and qualities, and also define the level of re- configuration and operation of the system. It should be understood in the same way as a product design.

In process design, designers must define manufacturing methods, conditions, process procedures, layouts, and so on. Systematization of such knowledge is a key issue in manipulating the various realms of manufacturing knowledge (quality assurance, machining, assembly, environmental management) in a way that satisfies a great number of constraints.

Vast amounts of knowledge must be taken into account during process design, and most such activities are carried out by multiple groups of designers. In support of their activities, an intelligence design system would help to induce efficiency in knowledge sharing and communication.

2. Knowledge systematization in process design

2.1. Ontology

Ontology is originally a branch of metaphysics dealing with the nature of beings. In the knowledge engineering field, ontology can be defined as a knowledge base which records explicit explanations of concepts, and relations among the concepts. Nowadays, such an understanding of ontology, where considerations of efficiency imbues knowledge with functions of sharing, reusing, searching and acquiring, draws considerable attention as a method of knowledge systematization.

Some description languages [1] or formats [2], [3] are proposed for building ontology. Several researches also attempt to develop a unified logical theory for knowledge integration and reuse by using ontology.

SHADE project [4] is an example attempting a virtual-integration of distributed CAE systems by ontology.

Ontology supporting process design should provide a way of discourse in collaborative activities among designers including multiple contexts.

2.2. Knowledge handling in process design

From the perspective of process-design support, we consider that the following objectives would vitalize knowledge handling:

  1. Preserving design qualities (utilizing engineering knowledge, and skilled engineers know-how),
  2. Preserving and maintaining product costs and qualities with minimum consumption of resources throughout the product life cycle,
  3. Observing principals, guidelines, and procedures of process design activities, and
  4. Speeding design work.
Here, in order to enable (1), it is essential to record or extract design intentions, namely explicit explanations of the knowledge handling processes, on condition that a certain system efficiently supports client browsing of historical records and typical examples.

Regarding (2), an evaluating function of proximity between concepts is expected. Such a function would include a pre-process that provides knowledge interpretation to unify viewpoints among designers and product development stages.

(3) represents the difficulties in meeting multiple constrains which are regulated by common design rules. Usually, such rules are based on manufacturing policy or philosophy, such as JIT concepts. In order to avoid any failure in communication, some sort of computerized help is expected.

Herein, principals are fundamental policies, guidelines are rules and criteria, and procedures are standardized development steps and their derivatives, in process design activities. We have proposed an implicit framework for guidance, one which eases the difficulties inherent in meeting constrains by knowledge systematization [5].

The functions for knowledge-handling aids mentioned above, help us to visualize a computerized platform which would enable designers to derive the maximum concentration from their intellectual work. Consequently, (4) can be done.

2.3. Ontology structure before reviewing

Figure 1 shows the overview of the ontology before reviewing [5].

Process design ontology can be divided into three subsets: principal & guideline ontology, procedure ontology, and design output ontology.

Figure 2 illustrates the structure of process design ontology. The principal & guideline ontology forms a layered structure of concepts including concepts of principal, guideline, and evaluation/decision criteria. Each concept is connected to others by one or more relations, which are part-of and/or refer-to relations.

Procedure ontology also has a layered structure, in which the top layer consists of output concepts which define the derivatives of design activities, such as process drawing, process flow chart. Also, each individual output concept contains such properties as having-product-name, having- process #. These concepts are combined by order relations, which regulate the process design procedures. The second layer consists of activity concepts which define specific design activities included in each output concept. Activity concepts are connected with output concepts by part-of relations.

In the procedure ontology and guideline concepts in the principal & guideline ontology, the activity concepts are connected by refer-to relations. Consequently, they form an explicit-concept-model of principals, guidelines, and procedures in a process design.

The consequence of design work is that each concept within the design output ontology automatically emerges as an instance of the related output concept. Individual instances have a specific name label and also have value-slots containing attributes corresponding, in a one-to-one relationship, to properties of the related output concept. Instances and their related output concepts are connected by instance-of relations.

This facility eases decision-making and traces relations among the principal & guideline ontology and the procedure ontology. As the design work progresses, designers can refer to possible alternatives or sets of values.

As the design activities are completed, attributes are filled in each instance. These values become design intentions.

In this model, the instances with their inherent design intentions become parts of the process design ontology and can be "browsed" as design knowledge.

3. Reviewing architecture through prototyping

We have built an ontology prototype regarding soldering process design, and evaluated the effectiveness of our ontology framework from a knowledge sharing perspective.

The ontology prototype is referred from a client module of a knowledge retrieval tool called DA/KnowteBook [6] through a distributed computing environment. The server module of DA/KnowteBook serves the ontology, and HTML-files containing reference texts and pictures. The client module of DA/KnowteBook provides a help to capture design intentions equivalent for the design output ontology.

Figure 3 and Figure 4 are the over view of the ontology prototype, and the system, respectively.

We have tried to design PWA (Printed Wiring Assembly) by referring concepts in Figure 3.

"PJ-XXX" in Figure 3 is the newly acquired knowledge, which was derived from our trial, composed of referred concepts, issued queries, new ideas, and comments.

From the aspect of daily operation, we have acknowledged that "intellectual byproducts" of design activities are well-organized and saved into value-slots of "PJ-XXX" herein, and then become available for reuse as knowledge. Moreover, in collaborative activities within two or more designers sharing the same perspective, the ontology can ease mining of such knowledge [6], by trucking relations between concepts and value-slots in concepts.

On the other hand, in the cases of collaborations sharing multiple perspectives, several issues have been addressed at our ontology framework.

For instance, contents of the activity "Defining Parts Layouts" in Figure 3 described from "To Avoid "Bridge"" context. This is a process-design-perspective. A product-design- perspective possibly requires "Defining Parts Layouts" to be also described from "Make Products Smaller" context.

In such a case, enabling extraction of "mission specific"contexts, or more specifically, contents from ontology containing heterogeneous perspectives is essential to maintain the efficiency in collaborative activities.

In addition, to hide residual viewpoints having no direct relations to confronted mission domains, and at the same time, to realize a navigating mechanism keeping appropriate levels among overall viewpoints, provide further help in a widely distributed collaborating framework.

  1. Refining ontology framework
Instead of the difficulty of multiple-perspective handling by a "single" ontology discussed above, such a complete and- large-scale ontology may contain enormous level of complicity, and then loose an important nature of ontology, knowledge visualization capability.

To overcome these difficulties in knowledge sharing among multiple disciplined activities, we have refined our ontology architecture at a conceptual level, shown in Figure 5.


The refined architecture contains several mission specific ontologies, which are defined or reconfigured independently. Such ontologies are not a complete set of systematized knowledge, but do not include any inconsistencies within the related areas of the mission domains.

Ontology Sharing & Contradiction I/F facilitates knowledge sharing from heterogeneous perspectives.

Herein, the system extracts sets of branches in ontology hierarchy, which consist of same concepts shared between ontologies, and both higher and lower layered concepts including relations. And then, the extracted branches are contradicted for several kinds of evaluations.

The same example with section 3 might make easier to understand the effects of such contradictions. A mission specific, product-design ontology, and process-design ontology shares the same concept "Defining Parts Layouts, " but also contains the concept including opposed aspects, like as "Make Products Smaller" and "To Avoid "Bridge", " so the system can navigate designers to reach the mutual satisfactory consequences avoiding soldering bridges.

To make this framework practical, many further researches are considered, especially analyzing and evaluating behaviors/features of cross sections between ontologies are essential and our main interest.

5. Conclusion

Efficiency in knowledge sharing and reuse means an aid for designers to maintain design qualities, and to speed design activities.

Knowledge systematization is a key to vitalize knowledge retrieval, especially in the cases where designers disciplined by different organizations or missions, collaborate through a distributed computing environment. However, many issues still remain to realize practical revel of knowledge sharing and reuse in distributed computing environments. Ontology is one of the possible alternatives to build such a knowledge retrieval framework into a virtual world. We would like to continue to focus on ontology from design support perspectives.

This research has been undertaken as a part of an IMS program.


[1] T. R. Gruber, "Ontolingua: A mechanism to support

portable ontologies", Technical Report KSL, Stanford

University, KSL, 1992, pp. 91-66.

[2] M. Geneserech and R. Fikes, "Knowledge Interchange

Format Version 2.2 Reference Manual", Logic Group

Report Logic-90-4, Stanford University, Computer Science

Department, 1991.

[3] Knowledge Based Systems, Inc., "IDEF5 Method Report",


[4] T. R. Gruber et al., "SHADE: Knowledge Based

Technology for the Re-Engineering Problem", ARPA, 1998.

[5] M. Chikano, "A Process Design System based on

Systematized Knowledge", PTK?98, Berlin, 1998.

[6] H. Takeda et al., "IMS GNOSIS Annual Reports", MSTC,

Japan, 1999.