Global Dynamics Network Construction from the Web

Abstract

In this paper, we will utilize patterns to query search engine in order to extract a large list of expressions and build their co-occurrence networks from the Web. Those expressions could be statistical terms like birth rates, events such as hurricanes, riots or facts as urbanization. Our method is based on specificities of the language in order to create patterns. Those patterns will help us to extract terms. The vast amount of available information, the power of search API engine like Yahoo or Google combined with our patterns allow us to extract with high accuracy expressions described above and find links between them.