CLARE : A System for Classificatory Knowledge Representation
A sound framework for classificatory knowledge representation (CLARE) is presented which is used for natural language understanding in connection with the system LINAS. In contrast to other approaches, a fundamental distinction is drawn between intensional and preextensional representations which are explicitly discerned as different layers in the knowledge base. Furthermore, a global classification by different dimensions is provided comprising the following ones: generality, collectivity, quantification, referentiality and factuality. Whereas these dimensions are bipolar (dichotomic or semi-continuous transition between two distinct poles) another hierarchically structured classification is added: a taxonomy of sorts, poly-hierarchically organized features and special means for lexical subcategorization.
Nutzung und Vervielfältigung:
Alle Rechte vorbehalten