The structure of research collaboration networks is of great relevance for the economic analysis of regional innovation and growth. Patent data are often utilized in this context since, notwithstanding their limitations, they provide useful insights on the relational aspects of innovative activities. While Social Network Analysis applied to patent data increasingly belongs to the methodological repertoire of regional economics, the potential contribution of communitydetection techniques to this field of enquiry remains largely unexplored. This work intends to fill this gap by showing that community detection techniques can be usefully applied to the analysis of co-patenting networks within studies of regional innovation and growth. We propose an original approach to assess the stability and robustness of the detected innovative communities. We apply different ensambles of methodologies on the same relational data and use statistical procedures to evaluate the level of agreement between the different procedures. The methodology is applied to a jointpatent application network drawn from the OECD REGPAT database for a sample of innovative firms operating in Italian technological districts. The identification of stable communities through statistical procedures is integrated with substantive information about the location, district participation and sectoral specialization of the firms in order to generate a taxonomy of local networks configurations.
- Social Network Analysis,
- Community Detection,
- Patent Networks,
- Innovation Networks
Available at: http://works.bepress.com/carlo_drago/95/