Information Cell Mixture Models: The Cognitive Representations of Vague Concepts

Abstract

Based on prototype theory for vague concept modelling, a transparent cognitive structure named information cell mixture model is proposed to represent the semantics of vague concept. An information cell mixture model on domain ${\it \Omega}$ is actually a set of weighted information cells L _i s, where each information cell L _i has a transparent cognitive structure ′L _i  = about P _i ′ which is mathematically formalized by a 3-tuple 〈P _i ,d _i ,δ _i 〉 comprising a prototype set $P_{i}(\subseteq \it \Omega)$, a distance function d _i on ${\it \Omega}$ and a density function δ _i on [0,+∞). A positive neighborhood function of the information cell mixture model is introduced in this paper to reflect the belief distribution of positive neighbors of the underlying concept. An information cellularization algorithm is also proposed to learn the information cell mixture model from training data set, which is a direct application of k-means and EM algorithms. This novel transparent cognitive structure of vague concept provides a powerful tool for information coarsening and concept modelling, and has potential application in uncertain reasoning and classification.