Structures of Semantic Networks: Similarities between Semantic Networks and Brain Networks

Abstract

Two networks were extracted from two large semantic networks, HowNet and synsets of WordNet, based on conceptual relations. Analysis of these networks shows that they are complex networks with features of small-world and scale-free. Results also show that semantic networks are similar to brain networks: (a) exponents of power law degree distributions are between 1.0 and 2.0, while exponents of brain function networks are around 2.0; (b) semantic networks have hierarchical structures while brain networks exhibit features of segregation and integration; (c) semantic networks are disassortative, in this not similar to brain function networks but similar to neural networks. Similarities between semantic networks and brain networks suggest that they may obey similar dynamic rules