Spread-gram: A spreading-activation schema of network structural learning

TL;DR

Spread-gram introduces a novel network embedding method inspired by human memory, leveraging spreading activation to better capture network structures, leading to improved analysis performance and faster convergence across diverse real-world networks.

Contribution

The paper presents a new network representation learning scheme based on spreading activation, addressing information bias and sparsity issues in traditional methods.

Findings

Significant improvement in multiple network analysis tasks.

Effective discovery of hierarchical network structures.

Fast training with linear complexity in edge number.

Abstract

Network representation learning has exploded recently. However, existing studies usually reconstruct networks as sequences or matrices, which may cause information bias or sparsity problem during model training. Inspired by a cognitive model of human memory, we propose a network representation learning scheme. In this scheme, we learn node embeddings by adjusting the proximity of nodes traversing the spreading structure of the network. Our proposed method shows a significant improvement in multiple analysis tasks based on various real-world networks, ranging from semantic networks to protein interaction networks, international trade networks, human behavior networks, etc. In particular, our model can effectively discover the hierarchical structures in networks. The well-organized model training speeds up the convergence to only a small number of iterations, and the training time is linear with respect to the edge numbers.