The centerpiece of the University of Maryland University Center project is our effort to model innovation networks in Maryland. This work builds on a conceptual framework and methodology developed by Scott Dempwolf as part of his doctoral dissertation research in which he created an innovation network model for Pennsylvania.
Innovation networks concerns the spatial and social dimensions of innovation, which is thought to be a key contributor to economic growth. Economic developers in particular are interested in the social, spatial, and technological characteristics associated with innovation and how to create regional environments that foster innovation and growth.
Innovation involves many different actors:
Each actor may be represented as a node in a network model
Innovation also involves many different types of relationships:
Each relationship between a pair of actors may be represented by a tie (line) in a network model. The combination of nodes and ties creates a network model.
Many of the actors involved are highly mobile (inventors and investors, for example). Other actors are strongly associated with specific places and may even be acting on behalf of that place (county economic developers for example). Therefore it makes sense to model places as nodes in the network, and to model the location of actors at specific points in time as ties (lines) between the actor nodes and the place nodes. This notion of embedding geography within a network framework is an important and novel part of this research.
When we use historical data to build such network models what we actually create is a kind of “archeological record” of innovation in a particular place. When we look at the innovation network models we are looking at actual history.
We may then use these models to explore regional innovation clusters that emerge within the network, or to see how certain actors are interconnected through multiple types of relationships. While geographic regions often tend to cluster together in the network we may also find key relationships with distant actors. Thus an important use of this tool is the exploration of previously unseen innovation networks.
We may also use Social Network Analysis (SNA) methods to measure different characteristics of network structure or the flows through the network. Interpreting the results through multiple theories of social interaction, behavior and communication can provide new insights into process of innovation. Thus a second use of the tool is to provide rich data models to advance social science research concerning the nature of the innovation process.
For example, when we compare measures of network structure and flow to measures of economic growth we discovered there are some important and significant relationships. These relationships shed new light on how innovation influences economic growth at the local level. This is a significant advance.
When we look at the networks over time we can also see the emergence and evolution of regional innovation clusters. Thus this research may also facilitate a better understanding of cluster dynamics by helping researchers pinpoint specific actors, events and relationships that were critical in cluster evolution. It will also help economic development practitioners and policy-makers see and understand regional innovation clusters from the perspective of innovation rather than the prevailing industry cluster method. This is a new perspective not currently available.
The implementation of this methodology in Maryland advances our research on network models of innovation and economic development. Where the Pennsylvania model is built on six types of relationships, we will add perhaps twenty more, including everything from incubators and federal labs to angel networks and professional associations. We are working with the Human-Computer Interaction Lab (HCIL) at UMD to create new visualizations and analytic tools for use by researchers, practitioners and policy makers. We will be focusing on developing tools for exploration and analysis, benchmarking and progress monitoring, and ultimately scenario modeling.
One addition to our approach that will be developed during the Maryland project is the integration of economic and employment data within the network context. Using county-level input-output data and the industry-occupation matrix, inter-industry transactions and employment by occupation can be modeled as a network. The integration of this industry and employment network with the innovation network faces some significant research challenges because the classification schemes for each are completely different and no good methods for translating one to the other currently exist.
A second addition that will be incorporated into the Maryland application is the integration of UMD’s Science and Technology Innovation Knowledge-Base (STICK) framework and methodology. STICK uses a multi-level data organization framework and advanced data mining techniques to model more subtle structures of ideas and influence in the innovation network through citations, publications, events and the like. STICK has the potential to integrate a measure of the intangible nature of innovation – the “buzz” so to speak – into the overall model.
In summary we are working on linking innovation, industries, the workforce and the “buzz” at the county level in a highly visual, interactive and dynamic network model, along with a suite of analytical tools that are clearly grounded in appropriate social science theory and a user interface that is accessible to researchers, practitioners and policy-makers interested in driving economic development through innovation.