Science overlay maps: a new tool for research policy and library management

Notice: This research summary and analysis were automatically generated using AI technology. For absolute accuracy, please refer to the [Original Paper Viewer] below or the Original ArXiv Source.

We present a novel approach to visually locate bodies of research within the sciences, both at each moment of time and dynamically. This article describes how this approach fits with other efforts to locally and globally map scientific outputs. We then show how these science overlay maps help benchmark, explore collaborations, and track temporal changes, using examples of universities, corporations, funding agencies, and research topics. We address conditions of application, with their advantages, downsides and limitations. Overlay maps especially help investigate the increasing number of scientific developments and organisations that do not fit within traditional disciplinary categories. We make these tools accessible to help researchers explore the ongoing socio-cognitive transformation of science and technology systems.

💡 Research Summary

The paper introduces “science overlay maps,” a visual analytics tool designed to locate bodies of research within the broader landscape of science both at a single point in time and dynamically over time. The authors begin by outlining the growing complexity of scientific knowledge and the limitations of traditional bibliometric indicators and static mapping techniques, which often fail to capture interdisciplinary and rapidly evolving research activities. They then describe how overlay maps build upon a global “base map” that represents the structural relationships among scientific fields. This base map is constructed from large bibliographic databases (e.g., Web of Science, Scopus) by clustering journals or subject categories based on citation and co‑citation patterns and projecting these clusters onto a two‑dimensional space using dimensionality‑reduction methods such as multidimensional scaling, t‑SNE, or UMAP.

In the overlay step, the user selects a specific set of publications—belonging to a university, corporation, funding agency, or thematic keyword set—and maps each item onto the base map according to its journal or subject classification. Visual attributes (color, size, opacity, animation) encode the identity of the entity, the volume of output, citation impact, and temporal evolution. This approach enables three primary applications: (1) benchmarking institutional research portfolios, highlighting strengths and gaps across disciplines; (2) visualizing collaboration patterns and the degree of cross‑disciplinary integration; and (3) tracking how policy or funding shifts influence the emergence and migration of research topics.

The authors illustrate the method with four case studies. First, they compare the research profiles of leading universities, showing a gradual shift from traditional engineering and natural sciences toward data science, artificial intelligence, and sustainability‑related fields. Second, they overlay corporate patent‑and‑paper data for firms such as IBM and Siemens, revealing the disciplinary focus of corporate R&D and the extent of interdisciplinary spill‑over. Third, they map the evolution of major funding programmes (e.g., EU Horizon, US NSF) over a decade, demonstrating how strategic priorities reshape the distribution of scholarly activity. Fourth, they create topic‑centric overlays for keywords like “nanotechnology” and “machine learning,” visualizing the diffusion of these technologies across disparate scientific domains.

The discussion balances the strengths of overlay maps—integrating quantitative metrics with intuitive visual cognition, supporting multi‑layer comparisons, and leveraging open‑source tools for reproducibility—against several limitations. The base map’s reliance on fixed classification schemes can lag behind emerging fields; journal‑centric mapping may underrepresent truly interdisciplinary work; data quality issues (missing or erroneous metadata) and publication delays can bias results; and increasingly complex visualizations demand specialized interpretive expertise. The authors also note computational challenges associated with real‑time animation and interactive dashboards.

In conclusion, overlay maps are positioned as a powerful instrument for exploring the socio‑cognitive transformation of science and technology systems, aiding research policy design, strategic planning, and library collection management. Future work is proposed in three areas: automating the update of base maps with contemporary classification systems, developing keyword‑driven mapping to better capture interdisciplinary research, and building user‑friendly interactive platforms that democratize access to these visual insights. By releasing the tools as open‑source resources, the authors aim to foster a collaborative ecosystem where scholars, policymakers, and information professionals can continuously refine and apply overlay mapping to the evolving landscape of global research.