Three aspects have been addressed to characterize agents’ innovative interactions (Lane, 2011): relational structures, shared processes and common functions. A new methodological approach that allows their investigation is here developed. Each of the aforementioned aspects is separately analyzed through the implementation of one of the following community detection methodologies. Respectively, these algorithms are: Clique Percolation Method (CPM), Infomap (IM), and Relevance Index (RI). Areas of co-existence of the three aspects are investigated by considering those intersections determined by groups of agents that are simultaneously detected together by each of the three methodologies. Finally, the implementation of a linear regression model is used to analyze which types of interactions are associated with larger size of these intersections. This final analysis demonstrates that the proposed methodology leads to the identification of areas of the system in which statistically significant elements emerge. The approach is implemented in a case study regarding a cycle of policy interventions, developed in Region Tuscany (Italy) from 2000 to 2006, aimed at supporting innovative network projects among local economic agents.