Shilling attacks pose a significant threat to the integrity and reliability of recommender systems by injecting fake user profiles to promote or demote targeted items. Existing generative approaches often suffer from unstable training dynamics and limited realism in the synthesized profiles. In this paper, we propose PGAN, a novel Penalized Generative Adversarial Network enhanced with latent space perturbations to generate high-quality, diverse, and undetectable shilling attack profiles. PGAN incorporates a gradient penalty to stabilize discriminator training and applies controlled noise perturbations in the generator’s latent space to improve robustness and attack diversity. We evaluate PGAN on real-world datasets and demonstrate that it consistently outperforms traditional statistical attacks and baseline GAN-based models across multiple evaluation metrics, including Hit Ratio@K, Prediction Shift, and attack success rate. Experimental results also confirm the realism of the generated profiles through similarity analysis with genuine users. Our proposed model outperforms traditional and state-of-the-art methods, achieving HR@10 scores of 0.2051 and 0.2076 on the MovieLens and Amazon datasets, respectively.

Large Language Models (LLMs) are increasingly integrated into Knowledge Graph (KG) construction and augmentation pipelines to reduce manual effort and enable scalable knowledge extraction, completion, and reasoning. While this integration offers substantial benefits, it also introduces new forms of bias and unreliability that extend beyond those observed in standalone LLMs or traditional knowledge graphs. In particular, biases originating from language models, such as social and representational bias, hallucination, prompt sensitivity, and domain coverage limitations that interact with structural and content biases inherent to knowledge graphs, result in compounded distortions that propagate across the pipeline. This paper provides a structured and comprehensive analysis of bias in LLM-augmented knowledge graph systems. We first review bias mechanisms in LLMs and standalone KGs, and then examine how these biases interact and amplify during key stages of LLM-based entity extraction, relation generation, graph completion, and reasoning. Based on this analysis, we introduce a unified taxonomy that characterizes bias as a pipeline-level phenomenon rather than an isolated model. We further consolidate recent evaluation metrics adapted for LLM-generated graphs, including semantic and soft lexical measures. Additionally, we survey representative datasets and benchmarks used to study bias in LLMs, KGs, and hybrid LLM–KG systems and identify open research gaps in developing pipeline-aware evaluation frameworks. This work aims to support the design of more reliable, accurate, and fair LLM-augmented knowledge graphs for engineering and domain-specific applications.

Shilling attacks pose a significant threat to the integrity and reliability of recommender systems by injecting fake user profiles to promote or demote targeted items. Existing generative approaches often suffer from unstable training dynamics and limited realism in the synthesized profiles. In this paper, we propose PGAN, a novel Penalized Generative Adversarial Network enhanced with latent space perturbations to generate high-quality, diverse, and undetectable shilling attack profiles. PGAN incorporates a gradient penalty to stabilize discriminator training and applies controlled noise perturbations in the generator’s latent space to improve robustness and attack diversity. We evaluate PGAN on real-world datasets and show that it consistently outperforms traditional statistical attacks and baseline GAN-based models across multiple evaluation metrics, such as Hit Ratio@K, Prediction Shift, and attack success rate. Experimental results also confirm the realism of the generated profiles through similarity analysis with genuine users. Our proposed model could surpass traditional and state-of-the-art methods, with HR@10 of 0.2051 and 0.2076 on MovieLens and Amazon datasets, respectively.