Through this design, CDNE incorporates local node proximity and mesoscopic community semantics without relying on an open-ended iterative procedure, and achieves strong performance in link prediction, cross-layer network reconstruction, and community-quality evaluation.
Abstract
Community structure provides important mesoscopic information for network representation learning, yet most community-aware embedding methods use detected communities as fixed contexts or auxiliary regularizers. This paper proposes CDNE, a Community-Driven Network Embedding framework based on two-stage community structure refinement. In the first stage, CDNE applies the Louvain algorithm to obtain initial community partitions and designs a community-guided random walk strategy to sample both topological neighbors and same-community nodes. The generated sequences are used to learn preliminary node embeddings with the skip-gram model. In the second stage, these preliminary embeddings are clustered to refine community assignments, and the refined communities guide a new round of community-aware embedding learning. Through this design, CDNE incorporates local node proximity and mesoscopic community semantics without relying on an open-ended iterative procedure. Experiments on multiple real-world networks show that CDNE achieves strong performance in link prediction, cross-layer network reconstruction, and community-quality evaluation. Ablation analyses further indicate that meaningful community partitions, community-guided walks, and second-stage refinement jointly contribute to the observed improvements, suggesting that community refinement is an effective strategy for learning discriminative node representations.
A combined HGT-based framework incorporating contrastive representation learning and deep clustering with multi-round training via pseudo-labels is presented, enabling the model to better deal with label scarcity, heterogeneous dependencies, and overlapping semantics in practical, complex, attributed networks.
Hamza Haddad, Hicham Attariuas, A. Younes· Edelweiss Applied Science an...· 0 citations
D2GSL constructs a semantic similarity channel and a spectral feature channel to model node relationships from both local semantic and global spectral views and introduces a hyperadjacency matrix that explicitly models inter-layer node correspondences and enables joint structural reconstruction across channels.
Jun-Chen Zhang, Xuhao Wei, Xiaolei Gu et al.· Computer Modeling in Enginee...· 0 citations
In this paper, for the first time, we study the community search problem over multimodal graphs. This task aims to identify a query vertex-containing subgraph that is both structurally cohesive and semantically coherent with multimodal query inputs (e.g., text and images). Existing community search methods fail to capture fine-grained multimodal semantics and do not support effective multimodal fusion. To address these limitations, we propose an adaptive community search framework ECHO, which includes two key components. (i) A Fine-grained Modality Extractor decomposes multimodal content into structured local semantic units to preserve details often lost in coarse representations, operating in an encoder-agnostic manner. (ii) A Dual-Track Mixture of Experts network decouples semantic and structural modeling into parallel tracks, utilizing a hierarchical MoE architecture for adaptive, query-aware feature fusion. Extensive experiments on real-world multimodal graphs demonstrate that ECHO consistently outperforms state-of-the-art methods in terms of community quality while achieving superior search efficiency.
Chengyang Luo, Zi-Xing Ding, Qing Liu et al.· Proceedings of the 32nd ACM...· 0 citations
LUCID, an LLM-guided, interpretable, training-free, and unsupervised community detection method, designed as a four-stage pipeline that achieves state-of-the-art performance and consistently outperforms leading unsupervised and semi-supervised baselines.
Aoting Zeng, Kai Wang, Jianwei Wang et al.· 0 citations
A generative model for temporal networks that jointly controls community evolution and dynamic node sets and is used as a benchmark to study the impact of the rate at which nodes join/leave the network on the performance of dynamic community detection algorithms.
Pei-Jie Zhong, Raul J. Mondragón, Richard G. Clegg· arXiv.org· 0 citations
Self-supervised learning on graphs is largely shaped by contrastive methods that depend on carefully designed augmentations, and by generative methods that reconstruct node attributes in the input space. Both paradigms can entangle representations with low-level input statistics rather than with relational structure. Joint-embedding predictive architectures (JEPA) instead learn by predicting latent targets rather than reconstructing inputs. Recent work has explored this idea for graph-level representation learning, but how to design JEPA-style objectives for node-level tasks, and which structural signals the predictor should condition on, remains less clear. We present NodeJEPA, a joint-embedding predictive architecture for node-level graph self-supervised learning. NodeJEPA masks structure-aware k-hop ego-subgraphs and trains a context encoder to predict the latent representations of the masked nodes. These targets come from an EMA-updated target encoder with stop-gradient. A structure-conditioned predictor integrates spectral and centrality descriptors through cross-attention. Variance, covariance, and Laplacian spectral regularizers help stabilize the embedding geometry, and an optional curriculum gradually increases masking difficulty during training. Because prediction occurs in latent space, NodeJEPA does not rely on input reconstruction or hand-crafted graph augmentations. We evaluate NodeJEPA on standard node classification benchmarks under linear probing and fine-tuning protocols, and conduct ablations on masking, prediction, and regularization design choices. Our study offers a practical recipe for node-level JEPA-style latent prediction on graphs, and clarifies when structural conditioning helps representation learning. Code, configurations, and evaluation scripts are publicly available at https://github.com/OliverZ-dot/Node-Jepa.
Tinghe Zhang, Jian Xu, Jiaheng Chen et al.· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
