Semi-automatic generation of a corpus of wikipedia articles on science and technology

Abstract

Despite the huge amount of scientific and technological content available on the World Wide Web, most of it is closed behind paywalls, as with academic journals, or almost invisible, as with institutional repositories. Wikipedia can act as a chain-transfer agent, providing people with an accessible, organized structure containing both understandable content and links to original sources. In Wikipedia, categories are collaboratively created and thus become a folksonomy rather than a true taxonomy. Consequently, categories are not a reliable tool to identify topics' organization. In this paper we describe a semi-automatic method, based on random walks, for determining a subset of pages containing scientific and technological content in the Spanish Wikipedia. Using the Unesco taxonomy, we determined the underlying graph structure of our corpus and detected clusters of pages strongly linked, establishing relationships between knowledge domains. Finally, we present the distribution of Wikipedia articles according to the Unesco taxonomy and the resulting map of scientific and technological content.