CoreCluster: A Degeneracy Based Graph Clustering Framework

Authors: Christos Giatsidis, Fragkiskos Malliaros, Dimitrios Thilikos, Michalis Vazirgiannis

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental Evaluation Here we present the experimental results of our framework on both the amelioration of the execution time and the quality of the clustering results.
Researcher Affiliation Academia Christos Giatsidis Ecole Polytechnique giatsidis@lix.polytechnique.fr Fragkiskos D. Malliaros Ecole Polytechnique fmalliaros@lix.polytechnique.fr Dimitrios M. Thilikos CNRS, LIRMM and Uo A sedthilk@thilikos.info Michalis Vazirgiannis Ecole Polytechnique and AUEB mvazirg@lix.polytechnique.fr
Pseudocode Yes Procedure CORECLUSTER(G). Input: A graph G. Output: A partition of V (G) into clusters. 1. k := δ (G). 2. q := 0. 3. Let Vk, . . . , V0 be the core expansion sequence of G. 4. For i = 0, . . . , k, let Gi be the i-core of G, 5. Let Sk = Vk. 6. Let Ak = {Ck 1 , . . . , Ck k} = Cluster(G[Sk]). 7. for i = k 1 to 0 do 8. Si =Select(Gi, Ak [ . . . [ Ai+1, Vi), 9. let Ai = (Ci 1, . . . , Ci i) = Cluster(G[Si]). 10. Return Ak [ [ A0.
Open Source Code No The paper refers to 'Supplemental material: Core Cluster: A degeneracy based graph clustering framework.' at http://www.lix.polytechnique.fr/dascim/wp-content/uploads/papers/corecluster14_supplemental.pdf, but does not explicitly state that source code for their methodology is provided there or at another location.
Open Datasets Yes We exploit the graph generator proposed in (Lancichinetti, Fortunato, and Radicchi 2008) to produce graphs with ground truth clustering structure. ... We also perform evaluations to a subset of the Facebook dataset (Traud, Mucha, and Porter 2011).
Dataset Splits No The paper describes the datasets used (artificial and Facebook) and evaluation metrics (NMI, Conductance), but does not explicitly detail specific train/validation/test splits used for the experiments.
Hardware Specification No The paper mentions 'hardware limitation' but does not provide specific details about the CPU, GPU, memory, or other hardware used for the experiments.
Software Dependencies No The paper mentions using 'Ng-Jordan-Weiss spectral clustering algorithm' and 'k-means++', but does not specify any software dependencies with version numbers (e.g., Python, PyTorch, scikit-learn versions).
Experiment Setup Yes Based on parameter space exploration, the values for the parameters of the Select procedure are chosen to be a = 0.8 and β = 5 (this choice appears to work optimally in our experiments).