Hyperbolic node embedding for signed networks

Signed network embedding methods aim to learn vector representations of nodes in signed networks. However, existing algorithms only managed embed networks into low-dimensional Euclidean spaces whereas many intrinsic features are reported more suitable for non-Euclidean spaces. For instance, previous works did not consider the hierarchical structures networks, which is widely witnessed real-world In this work, we answer an open question that whether hyperbolic space a better choice accommodate and embeddings can preserve corresponding special characteristics. We also propose method based on structural balance theory Riemannian optimization, embeds Poincar\'e ball space. This enables our approach capture underlying hierarchy because it be seen as continuous tree. empirically compare against six Euclidean-based baselines three tasks seven datasets, results show effectiveness method.