GCC-Spam: Spam Detection via GAN, Contrastive Learning, and Character Similarity Networks

2025-07-19

AI Summary

To address the surge of spam text on the Internet and the scarcity of labeled training data, this paper proposes GCC-Spam—a unified end-to-end framework for spam text detection. It integrates three synergistic components: (1) a Generative Adversarial Network (GAN) to synthesize high-fidelity pseudo-samples, mitigating data insufficiency; (2) a character-level similarity network that explicitly models fine-grained orthographic and phonetic confusion patterns; and (3) contrastive learning to enlarge the margin between legitimate and spam texts in the latent space, enhancing model robustness. Evaluated on real-world datasets, GCC-Spam significantly outperforms state-of-the-art methods—achieving higher accuracy with only a small number of labeled examples—and demonstrates strong resilience against adversarial attacks involving character substitution. The core contribution lies in the principled integration of GAN-based data augmentation, character-level confusion modeling, and contrastive boundary optimization, enabling effective and robust spam detection under data-scarce and adversarial conditions.

Technology Category

• Computer Vision: Generative Adversarial Networks (GANs) for Vision
• Machine Learning: Adversarial Learning & Robustness
• Natural Language Processing: Generation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Large pretrained models with web data
• User Modeling, Personalization and Recommendation: Attacks and countermeasures in recommendation systems

Abstract

The exponential growth of spam text on the Internet necessitates robust detection mechanisms to mitigate risks such as information leakage and social instability. This work addresses two principal challenges: adversarial strategies employed by spammers and the scarcity of labeled data. We propose a novel spam-text detection framework GCC-Spam, which integrates three core innovations. First, a character similarity network captures orthographic and phonetic features to counter character-obfuscation attacks and furthermore produces sentence embeddings for downstream classification. Second, contrastive learning enhances discriminability by optimizing the latent-space distance between spam and normal texts. Third, a Generative Adversarial Network (GAN) generates realistic pseudo-spam samples to alleviate data scarcity while improving model robustness and classification accuracy. Extensive experiments on real-world datasets demonstrate that our model outperforms baseline approaches, achieving higher detection rates with significantly fewer labeled examples.

Problem

• Detect spam texts using GAN and contrastive learning
• Address adversarial strategies and data scarcity in spam detection
• Improve detection accuracy with character similarity networks

Innovation

• Character similarity network captures orthographic and phonetic features
• Contrastive learning optimizes latent-space distance for discriminability
• GAN generates pseudo-spam samples to alleviate data scarcity