Applying neural networks to knowledge representation and determination of its meaning

Abstract

Knowledge representation is one of the first challenges AI community was confronted with. To be applicable, knowledge representation techniques must be able not only to represent the knowledge, but also to provide means to determine its meaning. The proposed knowledge representation techniques solve the problem of meaning determination by naming, i.e. by describing the meaning of represented knowledge. These descriptions are provided by database, knowledge base, ontology designers, which give names to tables, fields, classes, properties, relationships, etc. An alternative approach to the problem of determining the meaning would be a neural network approach applied to knowledge representation in a natural language that does not use names, but semantic categories. In this paper we propose a Hierarchical Semantic Form (HSF), a modification of localist approach of connectionist model, which, together with Space of Universal Links (SOUL) algorithm, is capable of representing knowledge in a natural language and interpreting its meaning by using the semantic categories. © 2007 Springer-Verlag Berlin Heidelberg.