Abstract:The thermodynamic approach of concentration mapping is used to discover communities in the directional friendship network of LiveJournal users. We show that this Internet-based social network has a power-law region in degree distribution with exponent \gamma = 3.45. It is also a small-world network with high clustering of nodes. To study the community structure we simulate diffusion of a virtual substance immersed in such a network as in a multi-dimensional porous system. By analyzing concentration profiles at intermediate stage of the diffusion process the well-interconnected cliques of users can be identified as nodes with equal values of concentration.

What problem does this paper attempt to address?

### Problems the paper attempts to solve This paper aims to solve the problem of community discovery in complex networks, especially for the directional friendship network among LiveJournal users. Specifically, the author attempts to study and identify the community structure in the network through thermodynamic methods. The following is a detailed description of this problem: 1. **Analysis of network characteristics**: - The author first analyzes the basic structural characteristics of the LiveJournal network, including degree distribution, small - world property, and node clustering. Research shows that the LiveJournal network has a power - law degree distribution with an exponent \(\gamma = 3.45\), and it is a small - world network with a relatively high clustering coefficient among nodes. 2. **Community discovery method**: - In order to study the community structure, the author proposes a diffusion simulation method based on thermodynamics. This method identifies communities by simulating the diffusion process of virtual substances in complex networks. The specific steps are as follows: - Select an initiator node in the network and inject a certain amount of virtual substance (such as "ink") into it. - Simulate the diffusion process of the virtual substance in the network, following physical laws similar to those of multi - dimensional porous systems. - Analyze the concentration distribution at different nodes during the diffusion process, especially in the intermediate stage of the diffusion process, and identify node clusters with similar concentration values. These node clusters are the communities. 3. **Advantages of the method**: - Compared with traditional graph partitioning algorithms, this method does not need to explicitly solve the motion equations of each microscopic component, but reveals the network structure through macroscopic phenomena (such as concentration distribution). - This method can effectively identify significant communities in the network, rather than simply decomposing the entire network into the smallest clusters. - The time complexity of the method is \(O(N\ln N)\), making it one of the fastest community discovery algorithms at present. ### Summary The main goal of this paper is to solve the community discovery problem in complex networks (especially the LiveJournal social network) by introducing a diffusion simulation method based on thermodynamics. This method can not only effectively identify the community structure in the network but also has high computational efficiency.

Thermodynamic approach for community discovering within the complex networks: LiveJournal study

Inferences in a Virtual Community: Demography, User Preferences, and Network Topology

A New Dynamic Community Model For Social Networks

Identify the Diversity of Mesoscopic Structures in Networks: A Mixed Random Walk Approach

Community structure in social and biological networks

Characterizing the community structure of complex networks

Dynamic Community Discovery Via Common Subspace Projection

Social significance of community structure: Statistical view

Think Locally, Act Locally: The Detection of Small, Medium-Sized, and Large Communities in Large Networks

Finding Communities in Linear Time: A Physics Approach

The Computer Science and Physics of Community Detection: Landscapes, Phase Transitions, and Hardness

Community analysis in social networks

Fundamental dynamics of popularity-similarity trajectories in real networks

A Survey on Theoretical Advances of Community Detection in Networks

Communities and beyond: mesoscopic analysis of a large social network with complementary methods

Time Is of the Essence: Analyzing the Effect of Vertex-Joining Time on Complex Network Evolution

Exploring Network Structure with the Density of States

Maps of random walks on complex networks reveal community structure

The statistical mechanics of Twitter

Laplacian Spectra of a Class of Small-World Networks and Their Applications

An Approach for Detecting Dynamic Communities in Social Networks