SDGNN: Learning Node Representation for Signed Directed Networks

Abstract

Network embedding is aimed at mapping nodes in a network into low-dimensional vector representations. Graph Neural Networks (GNNs) have received widespread attention and lead to state-of-the-art performance in learning node representations. However, most GNNs only work in unsigned networks, where only positive links exist. It is not trivial to transfer these models to signed directed networks, which are widely observed in the real world yet less studied. In this paper, we first review two fundamental sociological theories (i.e., status theory and balance theory) and conduct empirical studies on real-world datasets to analyze the social mechanism in signed directed networks. Guided by related socio- logical theories, we propose a novel Signed Directed Graph Neural Networks model named SDGNN to learn node embeddings for signed directed networks. The proposed model simultaneously reconstructs link signs, link directions, and signed directed triangles. We validate our model’s effectiveness on five real-world datasets, which are commonly used as the benchmark for signed network embeddings. Experiments demonstrate the proposed model outperforms existing models, including feature-based methods, network embedding methods, and several GNN methods.

Signed Graph Neural Networks: A Frequency Perspective

“…For these methods, we do not consider the sign of the links, since the signed edge information is not applicable for these methods. We also use the state-of-the-art GNN designs based on balance theory including SGCN (Derr et al, 2018), SNEA (Li et al, 2020), and SDGNN (Huang et al, 2021) for comparison.…”

“…We use SiNE (Wang et al, 2017), SLF (Xu et al, 2019), SGCN (Derr et al, 2018), SNEA (Li et al, 2020), and SDGNN (Huang et al, 2021) as baselines for comparison on link sign prediction tasks. We use a two layer GNN model along with a single hidden layer MLP classifier.…”

“…Different from these, our proposed methods are based on spectral domain analysis of signed graphs. SDGNN (Huang et al, 2021) is a recent work applicable to signed directed graphs based on balance and status theory. SSSNET (He et al, 2022a) is another GNN based work with a focus on clustering of signed graphs.…”

MSGNN: A Spectral Graph Neural Network Based on a Novel Magnetic Signed Laplacian

“…There is a demand for such tools because many important and interesting phenomena are naturally modeled as signed and/or directed graphs, i.e., graphs in which objects may have either positive or negative relationships, and/or in which such relationships are not necessarily symmetric [1]. For example, in the analysis of social networks, positive and negative edges could model friendship or enmity, and directional information could model the influence of one person on another [2,3]. Signed/directed networks also arise when analyzing time-series data with lead-lag relationships [4], detecting influential groups in social networks [5], and computing rankings from pairwise comparisons [6].…”

“…The goal of this paper is to introduce a novel Laplacian and an associated spectral GNN for signed directed graphs. While spatial GNNs exist, such as SDGNN [3], SiGAT [10], SNEA [11], and SSSNET [12] proposed for signed (and possibly directed) networks, this is one of the first works to propose a spectral GNN for such networks.…”