A Simple-Minded Network Model with Context-like Objects

Abstract

The aim of this paper is to describe a simple extension of semantic nets. In this formulation we have labelled nodes with directed arcs, but the directed arcs can lead to other arcs as well as nodes. In this model contexts are not differentiated as special objects, but rather that some nodes to a greater or lesser extent have roles as encoders of contextual information. This formulation is shown to be expressive enough to capture several aspects of context, namely: context-dependent inference, context specific learning, the selection of a relevant context and the generalisation of knowledge. Its strengths are its simplicity, the fact that it can relate and integrate several aspects of context and its connections with formal logic. It is not claimed that this is a model of any type of context found in human activity.