A Semantic Mapping of Food Systems Research and Innovation in Horizon Europe: Insights from FOOD2030 Pathways and Food Systems Thematic Clustering

As the first strategic plan of Horizon Europe (HE)—the European Union’s flagship Framework Program for Research and Innovation (R&I)— concluded in December 2024, it became essential to assess its alignment with key policy frameworks. One such framework is FOOD-2030, the EU’s research and innovation policy initiative designed to guide the transformation towards sustainable, healthy and inclusive food systems via existing key action pathways. FOOD-2030 promotes a systemic, multi-actor approach to address interconnected challenges such as climate change, biodiversity loss, resource scarcity, malnutrition, and food system resilience—while supporting broader strategies like the European Green Deal and Farm to Fork. This thesis explores how the FOOD 2030 research and innovation priorities are reflected across the full breadth of Horizon Europe-funded projects—spanning all program pillars and thematic areas. A comprehensive corpus of Horizon Europe project abstracts, along with their associated call topics and titles, was initially retrieved. For this study, the dataset underwent a rigorous filtering process to ensure direct relevance to food systems R&I within Horizon Europe projects. Initially, projects’ abstracts were filtered to include only, Pillar II, Cluster 6 (Food, Bioeconomy, Natural Resources, Agriculture and Environment - Horizon-CL6), resulting in a subset of 574 relevant abstracts out of the 14750 projects initially retrieved from CORDIS – Horizon Europe projects data-dump files. A further refinement step excluded projects under Cluster 6's 'Biodiversity' call topics, unless their titles explicitly mentioned "food systems", which allowed for the inclusion of 2 specific biodiversity-related projects. This selective approach yielded a final corpus of 475 projects. After subsequent classification into one or more of the 11 FOOD2030 Pathways for Action, based on pathway-specific manually created keywords list, 436 projects remained to form the basis for the detailed semantic and network-based analysis. Finally, within each pathway, preprocessed terms (hereby called lemma) were manually classified into 8 thematic clusters: driver, outcome, supply chain, food environments, individual factor, knowledge, actor, and system (plus, the ‘None’ cluster, which is excluded from the analysis). This thematic categorization was intentionally designed to balance analytical depth with interpretative clarity—capturing essential components of food system complexity (as emphasized in both the FOOD2030 framework and broader systems thinking literature) without excessive fragmentation or oversimplification. The clusters reflect widely recognized dimensions in food systems analysis, providing a structured lens to examine how research projects linguistically and conceptually represent food systems’ multifaceted components. The analysis highlights both shared features—such as the widespread salience of sustainability-related challenges like climate change—and distinct emphases that differentiate pathways and European countries. These patterns are revealed through multiple analytical layers: lemma co-occurrence networks that visualize the proximity of lemma within pathways; thematic cluster analyses of between and within lemma co-occurrences; and country-term bipartite mappings that showcase the distribution of thematic specialization and lemma across European countries. In addition, a strategic diagram approach, inspired by scientific papers co-word analysis (Callon et al., 1991), was applied to each pathway’s thematic clusters. By plotting the centrality (importance within the overall lemma network) against density (internal coherence of each thematic cluster), this method enables a synthetic view of thematic clusters prominence and cohesion in each pathway.