“Visualizing” the CG Community

The project is motivated by an empirical observation: over the past years a member of our group repeatedly encountered co-authors of another member at different conferences where the other member was not in attendance. This prompted some speculation on our part: What is the average coauthor distance within the community? What would a co-authorship graph of the community look like? Who are the centers of this graph? What is its diameter?

We set out to study and visualize collaborative structures within the computational geometry community. In particular, we investigate the formation and persistence of cliques and larger clusters, as well as the role of individual researchers as bridges or central actors. To this end, we compute a range of graph-theoretic properties, including diameter, radius, center, and connectivity measures across different temporal snapshots, and make available through a visualization.

The idea of quantifying scholarly proximity through graph distance is well established, most prominently through the Erdős number [6]. The Erdős number of a researcher is defined as the length of the shortest co-authorship path to Paul Erdős, where direct co-authors have Erdős number $1$, their co-authors have Erdős number $2$, and so on. This concept illustrates how collaboration graphs can be used to measure relational distance and has inspired a variety of tools and datasets, such as csauthors.net [2], which provide searchable views of co-authorship networks in computer science, as well as the Mathematics Genealogy Project [3], and the Academic Family Tree [4]. Building on these ideas, our project focuses on the computational geometry community and extends beyond static distance measures by emphasizing temporal evolution, structural metrics, and interactive exploration.

A variety of interactive applications exist for exploring communities and networked data, many of which are proprietary or require paid access, like PURE. In contrast, our application is tailored specifically to the CG community and is based entirely on publicly available data. This allows for open exploration of the community and its underlying structures. Moreover, another key feature of our application is the ability to explore how the community evolves over time, revealing how relationships and activity patterns develop. A snapshot of the application can be seen in Figure 1.

In its current version (March 27, 2026), our web application only includes the SoCG data from dblp. However, it can be straightforwardly extended to include additional conferences and journals indexed by dblp. For example, we can readily incorporate CCCG and JoCG, for which a dataset is already available. Since EuroCG has no official proceedings, it is generally not indexed, apart from a few isolated exceptions. To capture the computational geometry community as comprehensively as possible, future work will focus on compiling a dedicated EuroCG dataset; however, this will necessarily be a gradual process, largely because not all books of abstracts are readily accessible.

In addition, we will provide ongoing support of basic features, such as allowing individuals to be anonymized if they prefer not to appear in our application, or merging multiple dblp identifiers corresponding to the same person, along with similar adjustments. Moreover, we will add a variety of interesting statistics over time to enrich the content. For example, whether parts of the CG community graph exhibits small-world characteristics.

Initially, we had wished to add an “affiliation” filter to visualize collaborations by institution. However, the architecture of the dblp database makes loading this data dynamically infeasible. We therefore leave this as a future goal.