Hanjo D. Boekhout,Frank W. Takes
Abstract:Cliques, groups of fully connected nodes in a network, are often used to study group dynamics of complex systems. In real-world settings, group dynamics often have a temporal component. For example, conference attendees moving from one group conversation to another. Recently, maximal clique enumeration methods have been introduced that add temporal (and frequency) constraints, to account for such phenomena. These methods enumerate so called (delta,gamma)-maximal cliques. In this work, we introduce an efficient (delta,gamma)-maximal clique enumeration algorithm, that extends gamma from a frequency constraint to a more versatile weighting constraint. Additionally, we introduce a definition of (delta,gamma)-cliques, that resolves a problem of existing definitions in the temporal domain. Our approach, which was inspired by a state-of-the-art two-phase approach, introduces a more efficient initial (stretching) phase. Specifically, we reduce the time complexity of this phase to be linear with respect to the number of temporal edges. Furthermore, we introduce a new approach to the second (bulking) phase, which allows us to efficiently prune search tree branches. Consequently, in experiments we observe speed-ups, often by several order of magnitude, on various (large) real-world datasets. Our algorithm vastly outperforms the existing state-of-the-art methods for temporal networks, while also extending applicability to weighted networks.
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
This paper aims to solve the problem of quickly enumerating maximal cliques in weighted temporal networks. Specifically, the author proposes an efficient algorithm to enumerate so - called (δ, γ)-maximal cliques and addresses several key issues in the practical applications of existing definitions and methods. The following are the core problems and solutions in the paper:
1. **Problem background**:
- **Research on group dynamics in complex systems**: Cliques, that is, sets of fully - connected nodes in the network, are often used to study the group dynamics of complex systems. However, in the real world, these group dynamics usually have a time component, for example, conference participants move from one group discussion to another.
- **Enumeration of maximal cliques in temporal networks**: In recent years, some methods have introduced temporal and frequency constraints to enumerate (δ, γ)-maximal cliques to better reflect this phenomenon.
2. **Limitations of existing methods**:
- **Time interval expansion problem**: Existing definitions allow the time interval to extend beyond the actual time - edge instance range, resulting in some maximal cliques actually being strict sub - graphs of other maximal cliques.
- **Insufficient flexibility of weight constraints**: Existing methods only consider γ as a frequency constraint and cannot handle more complex situations in weighted networks.
3. **Solutions proposed in the paper**:
- **Improved definition of (δ, γ)-maximal cliques**: The author redefines (δ, γ)-cliques so that the start and end of the time interval are determined by the actual time - edge instances, and γ represents the cumulative weight threshold, rather than a simple frequency threshold. This not only solves the time interval expansion problem but also applies to weighted networks.
- **Efficient enumeration algorithm**:
- **Stretching Phase**: By optimizing the stretching phase, the time complexity is reduced from O(γm) to O(m), thus significantly improving the efficiency.
- **Bulking Phase**: A new bulking phase method is introduced, which uses the calculated values to efficiently expand the node set and prune the search tree branches, further accelerating the algorithm.
4. **Experimental results**:
- Through experiments on multiple real - world data sets, it is proved that the new algorithm is superior to existing methods in both time and space complexity, especially performing well on large - scale networks, with a speed increase of several orders of magnitude.
### Summary
By proposing a new definition of (δ, γ)-maximal cliques and an efficient enumeration algorithm, this paper solves the problems of efficiency and applicability in the enumeration of maximal cliques in temporal networks, especially outstanding in weighted networks. This improvement not only improves the algorithm performance but also ensures the accuracy and practicality of the results.