John N. Pougué-Biyong,Renaud Lambiotte
Abstract:In this article, we consider the problem of community detection in signed networks. We propose SignedLouvain, an adaptation of the Louvain method to maximise signed modularity, efficiently taking advantage of the structure induced by signed relations. We begin by identifying the inherent limitations of applying the standard Louvain algorithm to signed networks, before introducing a novel variant specifically engineered to overcome these challenges. Through extensive experiments on real-world datasets, we demonstrate that the proposed method not only maintains the speed and scalability of its predecessor but also significantly enhances accuracy in detecting communities within signed networks.
Social and Information Networks,Data Analysis, Statistics and Probability,Physics and Society
What problem does this paper attempt to address?
### What problem does this paper attempt to solve?
This paper aims to solve the **community detection problem in signed networks**. Specifically, the author proposes a new method named **SignedLouvain** to improve the application of the traditional Louvain algorithm in signed networks.
#### 1. **Background and Problem Description**
- **Limitations of the Traditional Louvain Algorithm**: The Louvain algorithm is a widely - used community detection method. It identifies the community structure in a network by maximizing modularity. However, when applied to signed networks, the traditional Louvain algorithm has some limitations. A signed network contains not only positive relationships (such as friendship) but also negative relationships (such as hostility). These negative relationships make it difficult for the traditional Louvain algorithm to effectively identify communities because it mainly relies on positive connections to form communities.
- **Complexity of Signed Networks**: The negative edges in signed networks increase the existence of local optimal solutions, causing the algorithm to be easily trapped in sub - optimal solutions. For example, in some cases, even in the early stages of the optimization process, the algorithm may not be able to find a suitable community assignment because the presence of negative edges makes the change in modularity negative, thus preventing further optimization.
#### 2. **Proposed Method**
- **SignedLouvain**: To overcome these problems, the author proposes the SignedLouvain method. This method allows nodes to move at different distances in the positive and negative networks by introducing the concept of multi - layer networks. Specifically, a node can move within the community of its positive neighbors (one step), and it can also move within the community of nodes two steps away in the negative network. This method not only considers positive relationships but also makes full use of negative relationships, thus better adapting to the characteristics of signed networks.
#### 3. **Experimental Verification**
- **Performance Improvement**: Through experiments on real - world datasets, the author proves that SignedLouvain not only maintains the speed and scalability of the traditional Louvain algorithm but also significantly improves the accuracy of community detection. Especially when dealing with sparse networks or networks with a large number of negative edges, SignedLouvain shows better performance.
- **Comparative Experiment**: The author also compares SignedLouvain with the relaxed version of the Louvain algorithm (RelaxedLouvain). Although RelaxedLouvain performs better in some cases, its computational cost is high and it may lead to inconsistent community structures. In contrast, SignedLouvain finds a better balance between speed and accuracy.
#### 4. **Conclusion**
- This paper successfully solves the challenges of community detection in signed networks and proposes a new algorithm, SignedLouvain, which can improve the accuracy of community detection while maintaining high efficiency. This provides a new direction for future research, especially when dealing with complex network structures.
### Summary
The core problem of this paper is **how to effectively perform community detection in signed networks**, and it proposes an innovative solution, SignedLouvain. By combining positive and negative relationships to optimize modularity, it improves the accuracy and efficiency of community detection.