Towards Efficient Detection of Overlapping Communities in Massive Networks
Authors: Bing-Jie Sun, Huawei Shen, Jinhua Gao, Wentao Ouyang, Xueqi Cheng
AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive tests on synthetic and large scale real networks demonstrate that the proposed strategies speedup the community detection method based on Poisson model by 1 to 2 orders of magnitudes, while achieving comparable accuracy at community detection. |
| Researcher Affiliation | Academia | 1CAS Key Laboratory of Network Data Science and Technology, Institute of Computing Technology, Chinese Academy of Sciences, Beijing, China 2University of Chinese Academy of Sciences, Beijing, China |
| Pseudocode | Yes | Algorithm 1: PARAMUPDATE: Updating the membership of connected nodes; Algorithm 2: DIMENOPT: Dimension level acceleration; Algorithm 3: LCNODE: Marking the dimension labels of converged Nodes; Algorithm 4: MAIN: Fast Poisson model algorithm |
| Open Source Code | No | The paper does not provide any explicit statement or link to the open-source code for the described methodology. |
| Open Datasets | Yes | We adopt both synthetic networks generated by LFR benchmark tool (Lancichinetti and Fortunato 2009) and a range of real-world networks (Yang and Leskovec 2015) to evaluate the performance of our proposed method.we adopt Amazon dataset, DBLP dataset, Youtube dataset, Live Journal dataset and Orkut dataset provided in the SNAP project for our experiments (Yang and Leskovec 2015). |
| Dataset Splits | No | The paper describes the generation of synthetic networks and lists real-world datasets but does not specify any training, validation, or test dataset splits. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used to run the experiments, such as CPU/GPU models or memory specifications. |
| Software Dependencies | No | The paper does not provide specific software dependencies with version numbers needed to replicate the experiment. |
| Experiment Setup | Yes | Three key parameters have to be set when generating benchmark networks, i.e., the power exponent of the degree distribution α, the power exponent of the community size distribution β, and the mixing parameter μ. The mixing parameter controls the fraction of a node’s links that connect to nodes in other communities. We vary the value of mixing parameter μ from 0.1 to 0.5 (the networks tend to be random graph when μ > 0.5) and α and β are set to be 2 and 1 respectively. We set the maximum degree as kmax = n 100 and the average degree k = 0.4 kmax, where n is the number of nodes. For community size, we set the minimum and maximum community size to be [ n 50, n 250]. The number of nodes is set to be 5,000 for our experiments. We require 10% of the nodes to belong to 2 communities. |