Hypergraph Change Point Detection using Adapted Cardinality-Based Gadgets: Applications in Dynamic Legal Structures

Hiroki Matsumoto,Takahiro Yoshida,Ryoma Kondo,Ryohei Hisano

2024-10-14

Abstract:Hypergraphs provide a robust framework for modeling complex systems with higher-order interactions. However, analyzing them in dynamic settings presents significant computational challenges. To address this, we introduce a novel method that adapts the cardinality-based gadget to convert hypergraphs into strongly connected weighted directed graphs, complemented by a symmetrized combinatorial Laplacian. We demonstrate that the harmonic mean of the conductance and edge expansion of the original hypergraph can be upper-bounded by the conductance of the transformed directed graph, effectively preserving crucial cut information. Additionally, we analyze how the resulting Laplacian relates to that derived from the star expansion. Our approach was validated through change point detection experiments on both synthetic and real datasets, showing superior performance over clique and star expansions in maintaining spectral information in dynamic settings. Finally, we applied our method to analyze a dynamic legal hypergraph constructed from extensive United States court opinion data.

Social and Information Networks

What problem does this paper attempt to address?

This paper attempts to solve the computational complexity problems encountered in analyzing hypergraphs in dynamic environments, especially in change point detection. Specifically, the author proposes a new method to convert hypergraphs into strongly - connected weighted directed graphs by adapting cardinality - based gadgets (CB - gadgets) and combining with symmetrized combinatorial Laplacian, thus effectively preserving the key cut information of the original hypergraph.

### Main Problems and Solutions

1. **Computational Complexity of Dynamic Hypergraph Analysis**:
- **Problem**: Traditional hypergraph analysis methods face huge computational challenges when dealing with dynamic data, especially when spectral information needs to be maintained.
- **Solution**: An adapted CB - gadget is introduced to convert hypergraphs into strongly - connected weighted directed graphs, enabling more efficient analysis in dynamic environments.

2. **Spectral Information Preservation in Change Point Detection**:
- **Problem**: How to effectively preserve spectral information is a key problem when performing change point detection in dynamic networks.
- **Solution**: By proving that the conductance of the transformed directed graph can be used as an upper bound of the harmonic mean of the conductance and edge expansion of the original hypergraph, the preservation of key cut information is ensured.

3. **Comparison with Existing Methods**:
- **Problem**: Existing methods based on clique expansion and star expansion perform poorly in dynamic environments.
- **Solution**: Experiments show that the proposed adapted CB - gadget method outperforms existing clique expansion and star expansion methods on both synthetic and real - world datasets.

### Formula Presentation

- **Hypergraph Conductance**:
\[
\phi_H(S)=\frac{\text{cut}_H(S)}{\min\{\text{vol}_H(S),\text{vol}_H(V\setminus S)\}}
\]
where,
\[
\text{cut}_H(S)=\sum_{e\in E}f_e(e\cap S)
\]
\[
\text{vol}_H(S)=\sum_{i\in S}d_i
\]

- **Harmonic Mean**:
\[
\mu_H(S) = 2\cdot\frac{\phi_H(S)\cdot\psi_H(S)/\beta}{\phi_H(S)+\psi_H(S)/\beta}
\]
where,
\[
\psi_H(S)=\frac{\text{cut}_H(S)}{\min\{|S|,|V\setminus S|\}}
\]

- **Main Theorem**:
\[
2\phi_G(T)\geq\mu_H(S)
\]
where \(T\subset\hat{V}\) and \(S = T\cap V\).

Through these formulas and methods, the author demonstrates the effectiveness of their method in preserving spectral information in dynamic environments and verifies its superior performance in change point detection tasks.