Construction and Analysis of Mongolian Word Co-occurrence Networks

In quantitative linguistics, the complex networks method is used to study the overall characteristics of language systems. This paper introduces the feature of the co-occurrence network of Mongolian. We constructed 41 Mongolian word co-occurrence networks based on five representative genre corpora, including Mongolian essays, poems, news reports, political articles, and laws. Statistical parameters of the networks are studied, including node number, edge number, diameter, average degree, average path length, clustering coefficient, and the degree distribution exponent. The results show that Mongolian word co-occurrence networks have the small-world effect and scale-free feature. In addition, there are significant differences in the networks constructed based on different language styles.