Explore collaboration in the Computational Geometry community through a dynamic co-authorship network built from open conference data provided by DBLP. The data is aggregated and organized into time-based snapshots, allowing you to discover how collaborative structures form, persist, and dissolve over time. Identify cliques and larger clusters, examine central and bridging researchers, and analyze graph-theoretic measures such as diameter, connectivity, and clustering to investigate whether the community exhibits small-world characteristics such as short collaboration paths combined with strong local structure. Inspired by concepts like the Erdös number, this app provides an interactive and evolving perspective on collaboration in computational geometry.
At launch, the application automatically retrieves all relevant data from dblp and computes several summary statistics. It also generates a visualization of the CG community graph for a randomly selected year. The web interface offers several tools for exploring the data in depth. Users can choose between displaying only co-author relationships or additionally showing the associated publications, and explore the data set interactively by clicking directly on any node or edge within the graph. This reveals detailed information associated with it, and provides even more options to dig into the data. A time-range slider allows the user to define a custom interval over which the data is aggregated and visualized. By pressing the "Play" button, the selected interval automatically advances year by year across the full timeline, dynamically updating the visualization to reflect the active data for the respective year(s).
Two coloring modes enhance interpretability:
Together, these features provide an intuitive and flexible way to explore the evolution of collaboration patterns in the CG community over time.
If you have trouble with or would simply like to reload the dataset, press this button:
To summarize the full story, the aggregated collaboration graph for ??‐?? contains researchers who collectively authored ?? papers, forming 204 connected components, including 82 singletons. The largest clique contains 14 researchers, whereas the average clique size is only 3.39, indicating that tightly interconnected groups remain relatively small. A similar pattern emerges when examining the connected components. Although the largest component is substantial, comprising 1,960 researchers and 2105 papers, and having a diameter of 12 and a radius of 6, the graph is otherwise highly fragmented. The mean component size is merely 11.7 nodes, increasing to 19.05 when singleton components are excluded. This suggests that, beyond the dominant core, collaboration is dispersed across many small components. Consistently, the average number of authors per paper is relatively modest, at 2.76 overall and rising to 3.13 when single-author papers are excluded, reflecting the prevalence of small collaborative teams.
This project was sparked by an empirical observation: a member of our group repeatedly encountered co-authors of another member across different conferences. This recurring experience suggested that co-authorship distances within the community may be small, and raised questions about the effective diameter of the collaboration graph, the location of its center, and the overall connectivity. Our visualization is not designed to rank or compare individuals for competitive purposes, but rather to support open and collaborative exploration and understanding of our community. Moreover, our intention is that it is "owned" and further developed by the community. Therefore, we invite everyone to explore the data, share insights, and contribute ideas or new features to shape its future.
The dblp computer science bibliography offers open bibliographic data (CC0 1.0) for major computer science journals and conference proceedings. Upon launch, your browser retrieves this information directly from dblp. All further interaction and processing takes place locally on your computer, using sql.js (MIT), graphology (MIT), and force-graph (MIT).
One last comment: With growing worldwide demand and a shrinking budget, dblp relies on community support. We kindly invite you to consider donating to Schloss Dagstuhl LZI.