Keywords Extraction from Chinese Document Based on Complex Network Theory

Keywords extraction is the process of choosing several words from a document to express its main idea. Keywords help people understand an article quickly and clearly. In recent years, more and more researchers pay attention to its research since its important role in text clustering, text classification, automatic abstracting, and text retrieval. This paper proposes an algorithm called EC-DC to extract keywords based on centrality measures of complex network. A document is mapped to a network with its words mapped to vertices and relations between words mapped to edges. Then, the importance of words is evaluated using eccentricity centrality and degree centrality. The most important K words are extracted as keywords. Experimental results show that the EC-DC algorithm has an improvement of about 9% in precision, recall and F-score compared to classical TFIDF algorithm.