Yuxin Wu
College of Engineering, Carnegie Mellon University, Moffett Field, CA 94035, USA.
*Corresponding author: Yuxin Wu
Abstract
A study was conducted on a user intent identification method for social bots based on graph attention networks, addressing the challenges of information manipulation and content abuse caused by automated accounts on social platforms. The method constructs a heterogeneous graph structure from users, content, and interaction relationships, integrating textual semantics, behavioral patterns, and network topology features. A multi-head attention mechanism is employed to achieve differentiated aggregation of neighborhood information. In the experimental design, publicly available international social media datasets were selected for comparative testing. The results demonstrate that the proposed method outperforms traditional text classification and graph convolutional models in multi-class intent recognition tasks, achieving higher accuracy and stability. It also exhibits stronger robustness in long-tail category detection and adversarial scenarios. The findings indicate that the graph attention network-based framework can effectively enhance the precision and interpretability of social bot intent identification, providing feasible technical support for platform governance and content regulation.
References
[1] Pires RP, Almeida AT. Interact2Vec — An efficient neural network-based model for simultaneously learning users and items embeddings in recommender systems. Appl Soft Comput. 2025;181:113408.
[2] Akhtar MM, Bhuiyan SN, Masood R, et al. BotSSCL: Social Bot Detection with Self-Supervised Contrastive Learning. Online Soc Netw Media. 2025;48:100318.
[3] Li L, Liu G, Liu Y, et al. A novel rumor detection method focusing on social psychology with graph attention network. Neuro-computing. 2025;626:129609.
[4] Chythanya KN, Bh. P, D. K, et al. A proposal of bidirectional grid long short term memory based model for user intention identification in on-line search query using text. Mater Today Proc. 2020;[prepublish].
[5] Ko HJ, Bae HJ, Hong D. Variable Impedance Control and Fuzzy Inference Based Identification of User Intension for Direct Teaching of a Mobile Robot. J Korean Soc Precis Eng. 2016;33(8):647-54.
[6] Wang S. A Cross-Distribution User Intention Identification Based on Topic Transfer. IOP Conf Ser: Mater Sci Eng. 2019;490(6).
[7] Brahma B, Kumari ALG, Panigrahi SB, et al. Interpretable Temporal-Spatial Graph Attention Network with Hyperbolic Sine Optimizer Algorithm for Alzheimer’s Disease Diagnosis Through Multiscale Feature Modeling. Biomed Mater Devices. 2025;[prepublish]:1-24.
[8] Luo X, Shu P, Liu N, et al. DG-MSGAT: A Biologically-informed Differential Gene Multi-Scale Graph Attention Network for predicting neoadjuvant therapy response in rectal cancer. Comput Methods Programs Biomed. 2025;271:108974.
How to cite this paper
Graph Attention Network-based User Intent Identification Method for Social Bots
How to cite this paper: Yuxin Wu. (2025) Graph Attention Network-based User Intent Identification Method for Social Bots. Advances in Computer and Communication, 6(4), 200-205.
DOI: http://dx.doi.org/10.26855/acc.2025.10.008