page 1  (8 pages)
2to next section

CERC Technical Report Series

Research Note

CERC-TR-RN-91-014

Percept Conceptualizations and its Knowledge Representation Schemes

Michael Sobolewski

October 1991

ACKNOWLEDGEMENT: This effort has been sponsored by Defense Advance Research Projects Agency (DARPA), under contract No. MDA972-88-C-0047 for DARPA Initiative in Concurrent Engineering(DICE).

Concurrent Engineering Research Center
West Virginia University
Drawer 2000, Morgantown WV 26506

2

Percept Conceptualizations and its Knowledge Representation Schemes*

Michael Sobolewski
Concurrent Engineering Research Center
West Virginia University
Drawer 2000, Morgantown WV 26506
email: [email protected]

Abstract

We suggest that relational conceptualizations are not adequate for representing knowledge expressed in natural languages at the computer level. First, one does not think explicitly in terms of functions and relations (however implicitly they exist). Moreover, one does think only in the terms of images of the real world, but not in the terms of the world objects explicitly. This image is determined by perception and is given by means of the senses. This means that image entities, called percepts, and real objects are different objects. Thus, natural language sentences can express only the image object's properties. A structure of percepts is defined in terms of the same primitives that we use to interpret natural sentences. In the percept conceptualization, functions and relations do not exist; only attributes and values exist as atomic conceptual primitives, and slots as molecular ones. A slot is a sequence of an attribute sequence (path) with a value at the last position. Elementary percept properties are built up from slots (one input slot and a set of output slots) and are considered as substitutes for simple sentences of natural language. Percept domains are defined and finally, the percept knowledge representation scheme is presented as a two level percept conceptualization.

* Acknowledgements - This work has been sponsored by Defense Advanced Research Project Agency (DARPA), under contract No. MDA972-88-C-0047 for DARPA Initiative in Concurrent Engineering (DICE)