A New Model to Compute The Information Content of Concepts Based on Cilin

Abstract

Information Content (IC) calculation plays an important role, because measuring similarities between words is a basic element of computational linguistics and artificial intelligence applications. This paper presents a novel IC value calculation model. Different from previous studies, the novel model considers the context characteristics of concepts in knowledge ontology, such as the depth of concepts, the number of sibling nodes or the number of leaf nodes of concepts, and introduces the dynamic parameter K to dynamically adjust the influence of depth factor on IC value. The experimental results based on PKU-500 dataset show that the correlation coefficient between human judgment and the calculation results of the new model is about 0.50, and compared with the existing IC calculation model, the accuracy of the similarity can be improved by 2% to 10%.