Kojaku et al., "Network community detection via neural embeddings" (2024)

Sadamori Kojaku, Filippo Radicchi, Yong-Yeol Ahn, and Santo Fortunato, Nature Communications 15(1), 9446 (2024)
DOI | arXiv | PDF | Code | Data

@article{kojaku2024network,
    author = {Sadamori Kojaku and Filippo Radicchi and Yong-Yeol Ahn and Santo Fortunato},
    title = {Network community detection via neural embeddings},
    journal = {Nature Communications},
    volume = {15},
    number = {1},
    pages = {9446},
    archivePrefix = {arXiv},
    eprint = {2306.13400},
    primaryClass = {physics.soc-ph},
    doi = {10.1038/s41467-024-52355-w},
    year = {2024},
}

Community detection with graph embedding, Community detectability

The paper connects node2vec to spectral embedding through the symmetric normalized Laplacian and derives its community-detection limit on stochastic block models. Node2vec reaches the information-theoretic detectability limit in the analysis, while simulations show strong performance on sparse and degree-heterogeneous networks. Davison et al. (2024) studies related guarantees for community detection with node2vec embeddings.