A Graph Neural Architecture Search Approach for Identifying Bots in Social Media

TL;DR

Bot detection on social platforms faces evolving adversaries and fixed-architecture limits. This work introduces a Graph Neural Architecture Search approach, $DFG$-$NAS$, adapted to Relational Graph Convolutional Networks to automatically optimize Propagation ($P$) and Transformation ($T$) steps while leveraging a heterogeneous graph of user metadata and follower relations. On TwiBot-20 (with $N=229{,}580$ nodes and $E=227{,}979$ edges), top NAS configurations achieve approximately $86$–$87\%$ accuracy and surpass state-of-the-art baselines, including BotRGCN and Botometer. The study demonstrates the value of NAS-driven neural network design in bot detection and points to broader applicability for dynamic social networks and cross-platform transferability.

Abstract

Social media platforms, including X, Facebook, and Instagram, host millions of daily users, giving rise to bots-automated programs disseminating misinformation and ideologies with tangible real-world consequences. While bot detection in platform X has been the area of many deep learning models with adequate results, most approaches neglect the graph structure of social media relationships and often rely on hand-engineered architectures. Our work introduces the implementation of a Neural Architecture Search (NAS) technique, namely Deep and Flexible Graph Neural Architecture Search (DFG-NAS), tailored to Relational Graph Convolutional Neural Networks (RGCNs) in the task of bot detection in platform X. Our model constructs a graph that incorporates both the user relationships and their metadata. Then, DFG-NAS is adapted to automatically search for the optimal configuration of Propagation and Transformation functions in the RGCNs. Our experiments are conducted on the TwiBot-20 dataset, constructing a graph with 229,580 nodes and 227,979 edges. We study the five architectures with the highest performance during the search and achieve an accuracy of 85.7%, surpassing state-of-the-art models. Our approach not only addresses the bot detection challenge but also advocates for the broader implementation of NAS models in neural network design automation.

Figures (3)

• Figure 1: Model used for Bot detection. User metadata is fed to the architecture proposed by NAS. The P step includes message aggregation from neighbour nodes. The T step includes the transformation process on each node based on neighbour relations. In the final part, an MLP decides whether the account belongs to a real user or a bot.
• Figure 2: Example of connections between the layers of NAS architecture. New T steps congregate information from all previous T steps. P steps propagate their embeddings and sum them up for the next T step.
• Figure 3: Permutations of Propagation (P) and Transformation (T) functions of the top-5 performing architectures from DFG-NAS. Their validation accuracies in the architecture search (from up to down) are: 87.01%, 86.99%, 86.95%, 86.89%, 86.82%