English

Dynamic Structural Clustering on Graphs

Data Structures and Algorithms 2021-08-27 v1

Abstract

Structural Clustering (DynCluDynClu) is one of the most popular graph clustering paradigms. In this paper, we consider StrCluStrClu under two commonly adapted similarities, namely Jaccard similarity and cosine similarity on a dynamic graph, G=V,EG = \langle V, E\rangle, subject to edge insertions and deletions (updates). The goal is to maintain certain information under updates, so that the StrCluStrClu clustering result on~GG can be retrieved in O(V+E)O(|V| + |E|) time, upon request. The state-of-the-art worst-case cost is O(V)O(|V|) per update; we improve this update-time bound significantly with the ρ\rho-approximate notion. Specifically, for a specified failure probability, δ\delta^*, and every sequence of MM updates (no need to know MM's value in advance), our algorithm, DynELMDynELM, achieves O(log2V+logVlogMδ)O(\log^2 |V| + \log |V| \cdot \log \frac{M}{\delta^*}) amortized cost for each update, at all times in linear space. Moreover, DynELMDynELM provides a provable "sandwich" guarantee on the clustering quality at all times after \emph{each update} with probability at least 1δ1 - \delta^*. We further develop DynELMDynELM into our ultimate algorithm, DynStrCluDynStrClu, which also supports cluster-group-by queries. Given QVQ\subseteq V, this puts the non-empty intersection of QQ and each StrCluStrClu cluster into a distinct group. DynStrCluDynStrClu not only achieves all the guarantees of DynELMDynELM, but also runs cluster-group-by queries in O(QlogV)O(|Q|\cdot \log |V|) time. We demonstrate the performance of our algorithms via extensive experiments, on 15 real datasets. Experimental results confirm that our algorithms are up to three orders of magnitude more efficient than state-of-the-art competitors, and still provide quality structural clustering results. Furthermore, we study the difference between the two similarities w.r.t. the quality of approximate clustering results.

Keywords

Cite

@article{arxiv.2108.11549,
  title  = {Dynamic Structural Clustering on Graphs},
  author = {Boyu Ruan and Junhao Gan and Hao Wu and Anthony Wirth},
  journal= {arXiv preprint arXiv:2108.11549},
  year   = {2021}
}
R2 v1 2026-06-24T05:25:42.677Z