Role-aware random walk for network embedding

Similarity (geometry) Representation
DOI: 10.1016/j.ins.2023.119765 Publication Date: 2023-10-11T04:39:13Z
ABSTRACT
Network embedding is a fundamental part of many network analysis tasks, including node classification and link prediction. The existing random walk-based methods aim to learn that preserves information on either proximity or structural similarity. However, the both role community important nodes. To address shortcomings methods, this paper proposes novel method for called RARE, which can be used different types networks even disconnected networks. proposed uses nodes preserve similarity in learned embeddings. walks generated through role-aware walk capture obtained are input Skip-gram model final In addition, RARE extended CRARE adds sampling high-order members customized so node's representation more network. performances evaluated multi-class classification, prediction, visualization tasks. Experimental results domain datasets indicate outperform baseline methods. further accelerated using parallelization generation process.
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (43)
CITATIONS (11)
EXTERNAL LINKS
PlumX Metrics
RECOMMENDATIONS
FAIR ASSESSMENT
Coming soon ....
JUPYTER LAB
Coming soon ....