R&I at MOST: Mapping Italian research networks on sustainable mobility funded under NextGenerationEU

Publication Type:

Conference Paper

Source:

Gerpisa colloquium, Shanghai (2026)

Abstract:

R&I at MOST: Mapping Italian research networks on sustainable mobility funded under NextGenerationEU

by Margherita Russo1, Fabrizio Alboni2, Pasquale Pavone3, 1, Simone Righi4,1

1 CAPP - Centro di Analisi delle Politiche Pubbliche, Università di Modena e Reggio Emilia, Italia

2 Università di Bologna

3 Università Pegaso, Italia

4 Dipartimento di Economia Marco Biagi, Università di Modena e Reggio Emilia, Italia

ABSTRACT

Research question (purpose)

Focusing on the flagship program “National Champions on Sustainable Mobility MOST”, the paper analyses the system-level investments in research and innovation on sustainable mobility, funded by the EU's Next Generation EU and implemented in the Italian National Recovery and Resilience Plan (NRRP) under the Mission 4, Component (M4C2). These investments aim to strengthen the research and technology transfer system by creating a public–private intermediary infrastructure that operates alongside existing regional policies. Started in 2023, the intermediary infrastructures and the research projects must be concluded by 2026. The extraordinary amount of public resources, about 1.7 million €, devoted to R&I in the MOST investment programme represents a unique opportunity to orient and enhance sustainable mobility in Italy.

Aiming to understand what might be leveraged from these resources when the programme is completed, the paper addresses three main research questions: those concerning the agents involved and the specific areas emerging from the R&I activities.

A preliminary research question (RQ1) concerns who are the agents involved in the governance of the 14 innovation intermediaries (spokes) that have been created under the MOST hub and in the R&I projects funded by the spokes[1]: which are their structural characteristics (public/private characteristic, sector of activity, size, location), which role they have in the governance of the spokes and the hub, and in the projects, which is their relative importance in terms of centrality in the governance and in the project networks across the policy programme.

The complementary RQ2 concerns which specific technological and social domains the R&I projects pursue in the general area of sustainable mobility.

Building on the findings from answering the first two research questions, the third research question (RQ3) integrates those results to map governance and project networks across thematic areas and to explore agents' research connections beyond the MOST hub within the other R&I investment programmes of M4C2.

Data and methodology (design)

Using an original database created ad hoc by the research team, in collaboration with a group of internship students of the Department of Economics Marco Biagi (University of Modena and Reggio Emilia, Italy), the study identifies the governance of the 14 spokes and, for each of them, the calls issued to implement R&I projects. For the 14 spokes and for the 156 research projects approved under the MOST investments, the data collection allows a detailed analysis of the 141 agents involved in the governance of the innovation intermediaries created to realize the R&I investment projects ‑ as proponent or participants of the MOST hub (50 agents), as leaders or affiliated members of the 14 spokes (55 agents), as project leaders or participants of the 156 projects (120 agents). In addition to identifying the agents involved in governance and in R&I projects, the database collects the textual descriptions of the goals and activities of each spoke and the themes of each call (for the M4C2 sub-measure 1.3-Partenariati Estesi, 1.4-Campioni Nazionali, 1.5-Ecosistemi dell’Innovazione). While the rules of the programme do not allow a member of a spoke to enter as a member of a project in the same spoke, multiple roles of the agents across spokes and R&I project ‑ as a spoke’s leader/affiliated member, or as a project’s leader/participant – creates a network of connection in the MOST hub, with 325 number of participations. Similar connections can be explored beyond this hub within the same sub-measure and across the other two sub-measures.

Text modelling was carried out using a four-stage methodology: vector representation of the texts (sentence embeddings), dimensionality reduction, document clustering, and extraction of characteristic words for each identified document group.[2]

A network analysis has been implemented to model two-mode and one-mode networks, characterising connections among agents and identifying the most central ones. Network statistics and agents’ centrality metrics contribute to characterising network structure across the thematic areas.

The main results (findings)

Through network and semantic analysis, the paper identifies key intermediaries, the bridging actors across investment lines, and the main thematic areas structuring collaboration. Information on the location of those agents allows us to outline the geography of the emerging competence network.

Ten thematic areas result from text analysis, in decreasing order of number of agents involved as project leader (in brackets): 7-Mobility, transport and urban infrastructure monitoring (26); 17-Thermal and fluid systems, control and testing of plant and vehicle systems (26); 18-Energy, hydrogen and storage/charging systems (17); 11-Design innovation and transfer to businesses and manufacturing sectors (4); 5-Computation, simulation and software/infrastructure development (3); 8-Coastal and hydraulic risk and local climate modelling (1); 9-Biodiversity, ecosystems and environmental resilience (1); 12-Communications, networks and integrated systems (1); 13-Advanced digital services, robotics and interaction (1); 15-Materials, additive manufacturing and the circular economy (1). Agents engaged in the top three thematic areas have some connections with the other spokes in the hub, while for most other R&I projects, the main connections are outside the hub within the same thematic area.

Measuring agents’ centrality using eigenvector metrics shows that the ranking of most central agents changes depending on their role: only in projects, only in governance, or in mixed roles.[3] Results show which Universities, research centres, and business companies are involved in those top positions. In addition,

Significance (practical and theoretical implications)

The paper contributes to advancing the debate on public–private intermediaries in innovation policy and offers early insights for the evaluation of an ongoing NRRP policy programme within the sustainable mobility domain and across the many thematic areas of the R&I policy programme. Such interconnections are relevant for leveraging future private and public investment in R&I.

 

[1] MOST - Centro Nazionale per la Mobilità Sostenibile; Spoke 1: Air mobility; Spoke 2: Sustainable road vehicle; Spoke 3: Watereways; Spoke 4: Rail transportation; Spoke 5: Light vehicle and active mobility; Spoke 6: Connected and Autonomous vehicle (CAV); Spoke 7: CCAM: connected networks and smart infra; Spoke 7: CCAM: connected networks and smart infrastructures; Spoke 8: MaaS and innovative services; Spoke 9: Urban mobility; Spoke 10: Logistics and freight; Spoke 11: Innovative materials and light weighting; Spoke 12: Innovative propulsion; Spoke 13: Electric traction system and batteries; Spoke 14: Hydrogen and new fuels.

[2] When using contextual embeddings for the semantic analysis of document corpora, it is important to clarify that these representations do not yield a ‘pure meaning’ of the text, separate from any other linguistic or formal component. On the contrary, the embedding synthesises the content of a text sequence as it is presented to the model, incorporating not only thematically relevant elements, but also recurring lexical signals, editorial formulas, structural labels and textual patterns that may be very frequent in the corpus. This aspect becomes particularly critical when texts are relatively short or semi-structured, as in the 156 calls, since under such conditions a few repeated terms can have a disproportionate impact on the vector representation and, consequently, on the proximity relationships between documents.

[3] This metric assigns each node a centrality score that is proportional not only to the number of direct connections (as calculated in the degree centrality index, but also to the importance of the nodes to which it is connected. In other words, being connected to central nodes contributes more significantly to a node’s centrality than connections to peripheral nodes. A high eigenvector centrality value, therefore, indicates a strategic position within the network, characterised by connections to nodes that are themselves influential.

  GIS Gerpisa / gerpisa.org
  4 Avenue des Sciences, 91190 Gif-sur-Yvette

Copyright© Gerpisa
Concéption Tommaso Pardi
Administration Alexandra Kuyo, Lorenza Monaco,, 

Powered by Drupal, an open source content management system
randomness