Automatic keywords extraction of Chinese document using small world structure

Small world structure characterized by short characteristic path length and high clustering coefficient is widely observed in many natural and man-made systems. Inspired by the small word structure found in English, we analyzed Chinese documents and construct cooccurrence networks to representing the correlations of words, where nodes are words and the cooccurrences of words in the same sentence make up links. The cooccurrence networks show obvious small world properties. By utilizing the impact of each node's absence on the characteristic path length of the network, we extract keywords, which can well represent the contents of the document.