SYMPOSIA PAPER Published: 01 January 1992
STP23778S

Materials Selection: Object-Oriented Structures for Factoring Polymer Information

Source

This paper discusses the fundamental issues involved in creating a taxonomy of engineering polymers capable of being implemented in class-instance, object-oriented programming languages. It reports the authors' experience of creating this taxonomy in POISE, our object-oriented database management system implemented in Smalltalk-80 [1]. Our account is centred on the issues involved in a strict factoring of the domain of polymers according to chemical composition, molecular scale and molecular arrangement, where these produce similar effects. Knowledge representation of materials information in our taxonomy is implicit and potentially far-reaching when rules of composition and scale are adhered to, rather than explicit and restricting when pragmatic groupings are created for particular purposes.

Knowledge representation of the relatively large scale micro-structures which create reinforced plastics pose particularly difficult issues which are discussed in detail. We find that the usual high-level, factoring of the materials domain into metals, ceramics, plastics and composites or into polymers and reinforced polymers is inappropriate for a taxonomy designed to represent plastics knowledge in a robust way independently of a user's particular purposes. This finding complements the authors' work [2] on the importance of separating perspectives for representing engineering design knowledge. A strict regime of independent perspectives is seen to be desirable at all levels of knowledge representation. This argues against the generality, or extensibility, of mixed taxonomies that are commonly seen to arise from a representation of materials information in a particular context, for a particular purpose.

POISE allows complex entities such as polypropylene, polyolefin, crystalline thermoplastic, and thermoplastic, which represent a line of direct ancestry in our taxonomy, to be stored and processed as largely independent computing objects. This accords nicely with the intuition that in deciding whether to make a washing-up bowl from polypropylene or polyethylene an engineer will have an idea of the generic properties of the two plastics and that this process is ill-served by a database model which searches through many individual grades of these polymers.

We comment on the principled support of independent browsing facilities which our taxonomy supports and argue for the nature and benefits of an object-oriented model of materials information which allows complex computing representations to evolve rather than to be designed, and which allows a representation of the structure of polymer knowledge to exist in a useful form independently of a program which provides particular services over the knowledge base.

Author Information

Demaid, A
The Open University, Walton Hall, Milton Keynes, U.K.
Ogden, S
The Open University, Walton Hall, Milton Keynes, U.K.
Zucker, J
The Open University, Walton Hall, Milton Keynes, U.K.
Price: $25.00
Contact Sales
Related
Reprints and Permissions
Reprints and copyright permissions can be requested through the
Copyright Clearance Center
Details
Developed by Committee: E49
Pages: 458–471
DOI: 10.1520/STP23778S
ISBN-EB: 978-0-8031-5200-7
ISBN-13: 978-0-8031-1437-1