Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Link Prediction for Social Networks using Representation Learning and Heuristic-based Features

Samarth Khanna, Sree Bhattacharyya, Sudipto Ghosh, Kushagra Agarwal, Asit Kumar Das

TL;DR

This work explores various feature extraction techniques to generate representations of nodes and edges in a social network that allow us to predict missing links and proposes combining heuristic-based features and learned representations that demonstrate improved performance for the link prediction task on social network datasets.

Abstract

The exponential growth in scale and relevance of social networks enable them to provide expansive insights. Predicting missing links in social networks efficiently can help in various modern-day business applications ranging from generating recommendations to influence analysis. Several categories of solutions exist for the same. Here, we explore various feature extraction techniques to generate representations of nodes and edges in a social network that allow us to predict missing links. We compare the results of using ten feature extraction techniques categorized across Structural embeddings, Neighborhood-based embeddings, Graph Neural Networks, and Graph Heuristics, followed by modeling with ensemble classifiers and custom Neural Networks. Further, we propose combining heuristic-based features and learned representations that demonstrate improved performance for the link prediction task on social network datasets. Using this method to generate accurate recommendations for many applications is a matter of further study that appears very promising. The code for all the experiments has been made public.

Link Prediction for Social Networks using Representation Learning and Heuristic-based Features

TL;DR

This work explores various feature extraction techniques to generate representations of nodes and edges in a social network that allow us to predict missing links and proposes combining heuristic-based features and learned representations that demonstrate improved performance for the link prediction task on social network datasets.

Abstract

The exponential growth in scale and relevance of social networks enable them to provide expansive insights. Predicting missing links in social networks efficiently can help in various modern-day business applications ranging from generating recommendations to influence analysis. Several categories of solutions exist for the same. Here, we explore various feature extraction techniques to generate representations of nodes and edges in a social network that allow us to predict missing links. We compare the results of using ten feature extraction techniques categorized across Structural embeddings, Neighborhood-based embeddings, Graph Neural Networks, and Graph Heuristics, followed by modeling with ensemble classifiers and custom Neural Networks. Further, we propose combining heuristic-based features and learned representations that demonstrate improved performance for the link prediction task on social network datasets. Using this method to generate accurate recommendations for many applications is a matter of further study that appears very promising. The code for all the experiments has been made public.
Paper Structure (17 sections, 3 equations, 2 figures, 4 tables)

This paper contains 17 sections, 3 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Data Used
  3. Methodology
  4. Data Preparation
  5. Feature Generation
  6. Structural Node Embeddings
  7. Neighborhood-based Node Embeddings
  8. Graph Neural Network-based Node Embeddings
  9. Heuristic-based Methods
  10. Combinations of different feature paradigms
  11. Classification
  12. Experimental results
  13. Initial Comparison of Algorithms
  14. Multiple Data Sets
  15. Deep Learning Architectures
  16. ...and 2 more sections

Figures (2)

  • Figure 1: Visualization of the workflow for adopted in our methodology for each data set.
  • Figure 2: A three-tower Neural Network Model as used in the 9th experiment. The layers depicted in blue process the WYS embedding for the source node, and the layers depicted in yellow process the WYS embedding for the destination node. These get combined with Heuristic features in the deeper layers of the network (depicted in green). Although this is not the best model for either data set, it performs well for all.