A Network-Specific Markov Random Field Approach to Community Detection
Authors: Dongxiao He, Xinxin You, Zhiyong Feng, Di Jin, Xue Yang, Weixiong Zhang
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We analyzed the new MRF-based method on several synthetic benchmarks and real-world networks showing its superior performance over the state of the art methods for community identification. |
| Researcher Affiliation | Academia | 1 School of Computer Science and Technology, Tianjin University, Tianjin, China 2 School of Computer Software, Tianjin University, Tianjin, China 3 College of Math and Computer Science, Jianghan University, Wuhan, China 4 Department of Computer Science and Engineering, Washington University, St Louis, MO, USA |
| Pseudocode | Yes | Algorithm 1: Efficient Max-Sum Belief Propagation |
| Open Source Code | No | The paper does not provide any explicit statements about releasing source code or links to a code repository. |
| Open Datasets | Yes | We used two types of synthetic networks: the Girven-Newman benchmarks (Girvan and Newman) and the LFR benchmarks (Lancichinetti, Fortunato, and Radicchi) with known community structures... The ten real-world networks that were analyzed (Newman; Xie, Kelley, and Szymanski; Sen et al.) are listed in Table 1... eight real-world networks with no known communities (Newman; Nelson, McEvoy, and Schreiber) which are listed in Table 3. |
| Dataset Splits | No | The paper describes the generation parameters for synthetic benchmarks and the characteristics of real-world networks used for evaluation. However, it does not specify explicit training, validation, or test dataset splits (e.g., percentages or sample counts) for their model's application, which is typical for community detection where the entire graph is often the subject of analysis. |
| Hardware Specification | No | The paper does not provide any specific details about the hardware used to run the experiments, such as CPU or GPU models, or memory specifications. |
| Software Dependencies | No | The paper does not specify any software dependencies with version numbers, such as programming languages, libraries, or frameworks used for implementation. |
| Experiment Setup | Yes | We ran each method 10 times and report the result with the highest objective... We used the programs of the existing methods from their authors and reported the results using their default parameters... When true communities are known we set K to the ground truth, whereas when communities are unknown we used Louvain method (Blondel et al.) to find K... For this benchmark [Girvan-Newman], each graph consists of 128 nodes divided into 4 groups of 32 nodes each. Each node has on average 16 in-edges... and z_out edges... For LFR Artificial Networks, we considered networks with 1000 nodes and the minimum community size c_min of 20 or 50. We varied the mixing parameter µ which specifies the fraction of the links of a node connecting to nodes outside of the node’s community from 0.1 to 0.5 with an increment of 0.05... The remaining parameters were kept fixed: the average degree d was set to 15, the maximum degree d_max to 0.5 * d, the maximum community size c_max to 0.2 * c_min, the exponent of power law distribution of node degrees τ to 2, and community size τ to 1. |