Clique Enumeration in Large Graphs

Clique enumeration seeks to list all complete subgraphs within a larger network, a task central to understanding cohesive structures in domains as diverse as social science, bioinformatics and cybersecurity. As clique enumeration is NP-complete, research has focused on methods that cope with scale and density. Classical techniques employ recursive backtracking—most notably the Bron-Kerbosch framework—enhanced by pivot selection and vertex ordering based on degeneracy to prune the search space. More recent advances introduce graph partitioning, hybrid strategies and parallelisation to distribute work across cores or clusters. These approaches balance memory footprint against computational overhead, leveraging sparsity metrics and dynamic pruning to avoid redundant exploration. Practical implementations integrate heuristics such as lowest-degree insertion and neighbourhood intersection counts to accelerate branch-and-bound. Applications span detection of tightly knit communities in social media, discovery of protein complexes in biological networks, identification of fraud rings in financial transactions and analysis of collaborative patterns in scholarly citation graphs. By revealing every maximal clique, these algorithms provide a granular view of network cohesion, informing downstream tasks in machine learning, statistical modelling and systems design.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Clique Enumeration in Large Graphs publication trend

The graph below shows the total number of articles in clique enumeration in large graphs across all publications each year (not limited to Nature Index journals).

Technical terms

Clique: A subset of vertices where every pair is connected by an edge.

Maximal clique: A clique that cannot be extended by adding any adjacent vertex.

Bron-Kerbosch algorithm: A principal recursive backtracking method for listing all maximal cliques.

Pivoting: A strategy in Bron-Kerbosch that chooses a vertex to reduce the number of recursive calls.

Degeneracy ordering: A vertex ordering by iteratively removing the lowest-degree vertex, bounding the enumeration complexity.

Graph partitioning: Dividing a graph into smaller subgraphs to enable parallel or distributed enumeration.

References

1. A linear delay algorithm for enumerating all connected induced subgraphs. BMC Bioinformatics (2019).
2. Efficient Maximal Clique Enumeration Over Graph Data. Data Science and Engineering (2016).
3. On Finding and Enumerating Maximal and Maximum k-Partite Cliques in k-Partite Graphs. Algorithms (2019).