Tutorial on NLP-Inspired Network Embedding

TL;DR

This tutorial reviews recent advances in network embedding techniques, focusing on methods that learn node representations in real-time for improved graph analysis tasks like link prediction.

Contribution

It provides an overview of recent online learning methods for network embedding, highlighting their applications and developments in the field.

Findings

DeepWalk, LINE, node2vec, struc2vec, megapath2vec enable effective online node embedding.

These methods improve real-time analysis of social networks and other graph data.

Recent developments facilitate better preservation of graph structure in embeddings.

Abstract

This tutorial covers a few recent papers in the field of network embedding. Network embedding is a collective term for techniques for mapping graph nodes to vectors of real numbers in a multidimensional space. To be useful, a good embedding should preserve the structure of the graph. The vectors can then be used as input to various network and graph analysis tasks, such as link prediction. The papers discussed develop methods for the online learning of such embeddings, and include DeepWalk, LINE, node2vec, struc2vec and megapath2vec. These new methods and developments in online learning of network embeddings have major applications for the analysis of graphs and networks, including online social networks.