Scale-Free Distribution in Chinese Semantic Field Network: A Main Cause of Using the Shortest Path Length for Representing Semantic Distance Between Terms

In large-scale network, using the weighted shortest path length to evaluate the relatedness between two terms is factually infeasible because of the actual time and space consumption, despite the fact that the related classic algorithm runs at complexity of \( o\left( {n^{3} } \right) \). However, in many natural language processing tasks, what we need to do is merely obtain the terms which are most related to a given term rather than obtain relatedness between every pair of terms, which makes it possible for using shortest path length between two terms within a large-scale complex network to evaluate semantic relatedness between two terms. Furthermore, one of the semantic field network’s important properties—scale-free distribution of node degree makes it much more feasible to use the shortest path length to evaluate semantic distance between two terms.