Co-word analysis is based on a sociological theory developed by the CSI and the SERPIA (Callon, Courtial, Turner, 1991) in the mid eighties. This method, originally dedicated to scientific fields, measures the association strength between terms in documents to reveal and visualise the evolution of scientific fields through the construction of clusters and strategic diagram. This method has since been successfully applied to investigate the structure of many scientific areas. Nowadays it occurs in many software systems which are used by companies to improve their business, and define their strategy but its relevance to this kind of application has not been proved yet. Using the example of economic and marketing information on DSL technologies from Reuters Business Briefing, this presentation gives an interpretation of co-word analysis for this kind of information. After an overview of the software we used (Sampler) and after an outline of the experimental protocol, we investigate and explain each step of the co-word analysis process: terminological extraction, computation of clusters and the strategic diagram. In particular, we explain the meaning of each parameter of the method: the choice of variables and similarity measures is discussed. Finally we try to give a global interpretation of the method in an economic context. Further studies will be added to this work in order to allow a generalisation of these results.
How to Cite:
Delecroix, B. and Eppstein, R., 2006. Co-word analysis for the non-scientific information example of Reuters Business Briefings. Data Science Journal, 3, pp.80–87. DOI: http://doi.org/10.2481/dsj.3.80