Semi-supervised network embedding with text information. (August 2020)

• Record Type: Journal Article
• Authors: Gong, Maoguo; Yao, Chuanyu; Xie, Yu; Xu, Mingliang
• Is Part Of: Pattern recognition. Volume 104(2020:Aug.)
• Publication Date: 2020-08
• Subjects: Network embedding -- Structure preserving -- Text representation -- Stacked auto-encoders; Pattern perception -- Periodicals; Perception des structures -- Périodiques; Patroonherkenning; 006.4
• DOI: 10.1016/j.patcog.2020.107347
• Languages: English
• ISSNs: 0031-3203

Abstract

Highlights: A semi-supervised method based on stacked auto-encoders for network embedding is presented. We explore the global structural information of the network by the structure preserving module and exploit the text features of nodes by the text representation module. A label indicator matrix and a supervised loss are proposed for the purpose of determining whether two nodes are in the same class and ensuring that the nodes in the same class have similar embedding vectors. Abstract: Network embedding plays a pivotal role in network analysis, due to the capability of encoding each node to a low-dimensional dense feature vector. However, most existing network embedding approaches only focus on preserving structural information in the network. The text features and category attributes of nodes are ignored, which are important to network analysis. In this paper, we propose an innovative semi-supervised network embedding (SNE) model integrating structural information, text features and category attributes into embedding vectors simultaneously. Specifically, we design a structure preserving module and a text representation module to capture the global structural information and the text features separately. Meanwhile, a label indicator matrix and a supervised loss are proposed for preserving category information and mapping nodes in the same class closer. We utilize stacked auto-encoders to explore the highly nonlinear characteristics of the network. By optimizing the reconstruction loss and the designed supervised loss jointly in the proposed semi-supervised model, the embedding vectors are finally learned. Extensive experiments on real-world datasets demonstrate that our method is superior to the state-of-the-art baselines in a variety of tasks, including visualization, node classification and clustering.