Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-view Learning for the Identification of Risky Users in Dynamic Social Networks

Francesco Benedetti, Antonio Pellicani, Gianvito Pio, Michelangelo Ceci

TL;DR

This work tackles risky-user identification in dynamic social networks by introducing T-SAIRUS, a temporally-aware, multi-view framework. It analyzes semantic content, network topology, and spatial proximity across overlapping temporal snapshots, fusing these views with an MLP and aggregating predictions via a time-weighted voting scheme. Experimental results on a real Twitter dataset show that incorporating temporal evolution and graph smoothing can improve detection of risky users compared to a static baseline, with performance depending on weighting and smoothing choices. The approach is computationally efficient, enabling incremental updates as new users are added, and demonstrates practical potential for online monitoring in social platforms.

Abstract

Technological progress in the last few decades has granted an increasing number of people access to social media platforms such as Facebook, X (formerly Twitter), and Instagram. Consequently, the potential risks associated with these services have also risen due to users exploiting these services for malicious purposes. The platforms have tools capable of detecting and blocking dangerous users, but they primarily focus on the content posted by users and usually overlook additional factors, such as the relationships among users. Another key aspect to consider is that users' beliefs and interests evolve over time. Therefore, a user who can be considered safe at one moment might later become malicious, and vice versa. This work describes a novel approach to node classification in temporal graphs, aimed at classifying users in social networks. The method was evaluated on a real-world scenario and was compared to a state-of-the-art system that treats the network as a static entity. Experiments showed that taking into account the temporal evolution of the network, in terms of node features and connections, is beneficial.

Multi-view Learning for the Identification of Risky Users in Dynamic Social Networks

TL;DR

This work tackles risky-user identification in dynamic social networks by introducing T-SAIRUS, a temporally-aware, multi-view framework. It analyzes semantic content, network topology, and spatial proximity across overlapping temporal snapshots, fusing these views with an MLP and aggregating predictions via a time-weighted voting scheme. Experimental results on a real Twitter dataset show that incorporating temporal evolution and graph smoothing can improve detection of risky users compared to a static baseline, with performance depending on weighting and smoothing choices. The approach is computationally efficient, enabling incremental updates as new users are added, and demonstrates practical potential for online monitoring in social platforms.

Abstract

Technological progress in the last few decades has granted an increasing number of people access to social media platforms such as Facebook, X (formerly Twitter), and Instagram. Consequently, the potential risks associated with these services have also risen due to users exploiting these services for malicious purposes. The platforms have tools capable of detecting and blocking dangerous users, but they primarily focus on the content posted by users and usually overlook additional factors, such as the relationships among users. Another key aspect to consider is that users' beliefs and interests evolve over time. Therefore, a user who can be considered safe at one moment might later become malicious, and vice versa. This work describes a novel approach to node classification in temporal graphs, aimed at classifying users in social networks. The method was evaluated on a real-world scenario and was compared to a state-of-the-art system that treats the network as a static entity. Experiments showed that taking into account the temporal evolution of the network, in terms of node features and connections, is beneficial.

Paper Structure

This paper contains 14 sections, 10 equations, 2 figures, 3 tables.

Table of Contents

  1. Introduction
  2. The method T-SAIRUS
  3. Semantic analysis of the posted content
  4. Topological analysis of the relationship network
  5. Spatial analysis of the closeness among users
  6. Model fusion
  7. Voting phase
  8. Graph smoothing
  9. Experiments
  10. Dataset
  11. Experimental setup
  12. Results and Discussion
  13. Conclusions
  14. Acknowledgments.

Figures (2)

  • Figure 1: T-SAIRUS framework. The figure emphasizes how the network relationships change over time, while the coloured bars below each user represent the score associated with them, which can also evolve.
  • Figure 2: Distribution and cumulative distribution of the number of tweets per month in the considered dataset.