Motifs and Motif Generalization in Chinese Word Networks.

The most significant semantic unit of Chinese language is words composed of individual characters. This com–positional structure produces great variability and representability compared to individual characters, which is quite distinct from other languages. In this paper we utilized complex networks to model the composition of words from characters. We focus on network motifs, the local pattern which appears more often in a statistically significant sense. Network motifs describe the most significant connection pattern between these nodes. We investigated their functions and semantical relationship. We also investigated different generalizations of network motifs and analyzed the larger pattern in the complex networks. As the word network is quite huge and the motif detection is very slow when motifs are much larger, for larger pattern in the network we used topology generalization of simple motifs rather than carry out a thorough motif detection task. The results on motifs and motif generalization in this paper not only offer us a big picture how Chinese words are formed, but also support the conclusion that motifs play a very important role in research of complex systems.