Quantifying impact and innovation in scientific publications: A structural–semantic framework combining co-authorship and research topics in Italian computer science
Year: 2026
Type: article
Abstract: Traditional bibliometric indicators often fail to capture the multidimensional nature of scientific contributions, reducing complex intellectual identities to simple numerical scores. This study introduces a structural-semantic framework designed to quantify impact and innovation by integrating collaboration patterns with research topics. We model the complete population of 1,229 Italian computer science scholars using a hybrid 896-dimensional representation space. Our methodology combines structural graph embeddings derived from co-authorship networks with semantic representations of publication abstracts, utilizing transformer-based models and density-based clustering. To enhance interpretability, we employ a two-stage pipeline using large language models to generate scientifically coherent labels for identified research themes. Our integrated approach captures academic identity, paving the way for the creation a quantitative measure of research similarity. The framework successfully distinguishes researcher roles—such as specialists and interdisciplinary connectors and identified over 100 potential collaboration opportunities between researchers with shared expertise but no prior history of working together. Longitudinal analysis revealed three distinct waves of innovation in the Italian computer science landscape over the past 40 years, shifting from foundational theoretical infrastructure to applied artificial intelligence and health informatics. Furthermore, the model showed high predictive validity in forecasting future research interests. This framework provides a robust tool for science policy and institutional planning, offering a more faithful representation of scientific evolution than traditional citation-based metrics.
Source: PLOS complex systems.
Institutions: Utrecht University, University of Genoa
Language: English
Subfield: Library and Information Sciences
Field: Social Sciences
Domain: Social Sciences
Open Access status: gold
Estimated APC paid: $3,043
Funders: European Commission, NextGenerationEU
Award: G1373274450
OpenAlex ID: W7237208591
