Page - (000181) - in Knowledge and Networks

175 transport.1 Beyond this highly technological core, the district was also given a sci- ence base through involvement of public universities and the national research insti- tute. The governance was a holding of industrial and public laboratories and institutes. This holding was managed by a dedicated administration whose role was to orchestrate collaborations to promote horizontal partnerships and encourage the recombination of knowledge bases. The explicit mission of the district, as knowl- edge integrator, was to intensify public-private collaborations and to put in place private-to-private interactions connecting firms from different industries within and outside the district.2 This cross-sectoral approach to baseline research was orchestrated by the admin- istration through active management of the network. Not only were projects were selected, but district members and external partners were called upon to participate in order to encourage the convergence between science and technology. IMAST management was able to attract an increasing number of firm leaders (the number doubled during the 10 years examined), and the science base was expanded to involve an increasing number of departments, universities, and institutes. In the fol- lowing section the R&D collaboration network among members is first defined and then analyzed by means of social network analysis. The data in this chapter refer to all IMAST R&D projects subsidized by both national and international grants. These projects can be used to study the structure of collaboration networks linking members among themselves and with the rest of the world.3 They also express IMAST’s policy on collaboration insofar as they were the direct result of the administration’s innovation strategy. The data encompass 24 R&D research projects undertaken by the district between 2006 and 2013. These project data constitute a two-mode network of organizations participating in projects. More formally, let N be the set of n TD’s members (asso- ciated members and external partners) and P be the set of the p R&D projects observed for the n members over time. An affiliation matrix A n p´( ) can be defined. The matrix entry aik is equal to 1 if the organisation i NÎ participates in the project k PÎ , and is equal to 0 otherwise. We use the conversion approach (Everett & Borgatti, 2013) to obtain an actor- by- actor adjacency matrix G from the two-mode network. In matrix G the entries are equal to the number of research projects shared by two organizations and 0 if two organisations have never collaborated in a research project. In order to highlight the structural changes that occurred over time, separate adjacency matrices were derived for each year (Prota & Vitale, 2014). Each of these temporal slides can be described as a graph G N L TT , ,( ), with T being the set of ordered time points t TÎ . The set of actors n NÎ and the links l LÎ change over time according to individual par- ticipation in the projects. 1 http://www.imast.biz/index.php?option=com_content&view=article&id=1&Itemid=2&lang=en 2 The governance structure, the rules, and the composition of the district were highlighted by the manager during an in-depth interview held in February 2014. In particular, the active role of man- agement was clarified by the manager and further discussed with members. 3 We thank the TD’s administrative staff who helped us update the data to March 2013. 9 Topology and Evolution of Collaboration Networks