Title :
The role of taxonomy in drawing interpretation
Author :
Pasternak, Boris
Author_Institution :
AI-Lab., Hamburg Univ., Germany
Abstract :
The objective of this paper is to demonstrate how taxonomic structures in knowledge representation can be exploited for drawing interpretation tasks. After a brief survey of the different levels of knowledge representation used by today´s interpretation systems, same basic assumptions about the nature of drawing interpretation and the representation of knowledge are discussed. Based on these widely valid assumptions, the advantages of an explicit declaration of the objects´ taxonomy and partonomy (part-of relations) are presented. In addition to the obvious results of a better readability and maintainability of the represented knowledge, it is shown how a complete and correct interpretation strategy can be automatically derived from only a handful of declarations. By this declarative approach the paper might shed some light on the special role of drawing interpretation as a mediator between the graphical and the applicational world. Taking into account this duality the paper introduces two different schemes of taxonomy and tries to establish rules for reasoning on dual taxonomies
Keywords :
document image processing; image recognition; knowledge representation; spatial reasoning; declarative approach; drawing interpretation; dual taxonomies; image recognition; knowledge maintainability; knowledge readability; knowledge representation; object partonomy; object taxonomy; part-of relations; reasoning; scanned drawings; taxonomic structures; taxonomy; Aggregates; Concrete; Engineering drawings; Functional analysis; Kernel; Knowledge representation; Specification languages; Taxonomy; Three dimensional displays; Writing;
Conference_Titel :
Document Analysis and Recognition, 1995., Proceedings of the Third International Conference on
Conference_Location :
Montreal, Que.
Print_ISBN :
0-8186-7128-9
DOI :
10.1109/ICDAR.1995.602022