Distributed scientific communication in the European information society: Some cases of 'Mode 2' fields of research

Can self-organization of scientific communication be specified by using literature-based indicators? In this study, we explore this question by applying entropy measures to typical “Mode-2” fields of knowledge production. We hypothesized these scientific systems to be developing from a self-organization of the interaction between cognitive and institutional levels: European subsidized research programs aim at creating an institutional network, while a cognitive reorganization is continuously ongoing at the scientific field level. The results indicate that the European system develops towards a stable level of distribution of cited references and title-words among the European member states. We suggested that this distribution could be a property of the emerging European system. In order to measure to degree of specialization with respect to the respective distributions of countries, cited references and title words, the mutual information among the three frequency distributions was calculated. The so-called transmission values informed us that the European system shows increasing levels of differentiation.

💡 Research Summary

The paper investigates whether the self‑organization of scientific communication in Europe can be captured through literature‑based indicators, focusing on “Mode 2” fields of knowledge production. Mode 2 denotes a context‑driven, interdisciplinary, and application‑oriented mode of research that is increasingly supported by European Union (EU) research programmes. The authors hypothesize that the emergence of a European scientific system results from the interaction of two levels: an institutional level, shaped by EU‑funded programmes that create a network of research institutions across member states, and a cognitive level, where the scientific fields themselves reorganize through the distribution of cited references and title‑words.

To test this hypothesis, the authors assembled a dataset of publications from three representative Mode 2 domains—biotechnology, information‑communication technologies, and environmental science—published between 1990 and 2000 by authors affiliated with the 12 EU member states. For each paper they extracted (i) the country of the author(s), (ii) the list of cited references, and (iii) the most salient words from the title. These data were aggregated to produce three frequency distributions: (a) the number of papers per country, (b) the occurrence of each cited reference across countries, and (c) the occurrence of each title‑word across countries.

The analytical framework draws on information theory. Shannon entropy (H) was computed for each distribution to quantify its uncertainty or dispersion. Joint entropies (H_xy) were then calculated for each pair of distributions (country‑reference, country‑title‑word, reference‑title‑word). Mutual information (I), also called transmission, was derived as I = H_x + H_y – H_xy. A higher transmission value indicates stronger dependence between the two variables, i.e., a greater degree of specialization of a country with respect to a particular set of references or title‑words.

The results reveal two complementary dynamics. First, the overall entropy of the reference and title‑word distributions declines over the decade, suggesting that the European system moves toward a more stable, less random allocation of knowledge artefacts among its members. Second, the transmission values increase steadily. The country‑reference transmission shows the most pronounced rise, indicating that EU research funding and collaborative frameworks have fostered the emergence of national “citation niches” where certain countries become central nodes in specific reference networks. The country‑title‑word transmission also grows, reflecting the formation of national linguistic or thematic signatures in the way research topics are labelled.

These findings support the authors’ hypothesis of a dual self‑organizing process. At the institutional level, EU programmes provide the scaffolding for cross‑border collaboration, aligning funding streams, infrastructure, and evaluation criteria. At the cognitive level, researchers respond to these incentives by converging on shared citation practices and terminology, while simultaneously differentiating along national lines as they develop specialized expertise. The combination yields a European scientific system that is both integrated (stable distribution of references and title‑words) and differentiated (increasing mutual information).

Methodologically, the study demonstrates that entropy‑based indicators can operationalize the abstract concept of “self‑organization” in scientific communication. For policymakers, monitoring transmission values offers a quantitative early‑warning system: rising values may signal successful specialization and the emergence of European research strengths, whereas stagnation could indicate a need to adjust funding mechanisms.

The paper concludes by recommending extensions of the approach to additional disciplines, longer time spans, and comparative analyses with non‑European regions. Such work would test the generality of the observed patterns and refine the use of information‑theoretic metrics as tools for science policy evaluation in the evolving knowledge‑based economy.