Deep Learning-Based Approaches for Detecting AI-Generated Fake Reviews in Online Platforms

Bangari, Ashutosh (2025) Deep Learning-Based Approaches for Detecting AI-Generated Fake Reviews in Online Platforms. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (775kB) | Preview
Preview	PDF (Configuration Manual) Download (530kB) | Preview

Abstract

The exponential growth of Artificial Intelligence and Natural Language Processing has blurred the boundaries between human-written and machine-generated text, creating new challenges in detecting synthetic content, ensuring authenticity, and maintaining academic integrity. While these technologies have advanced digital communication and sentiment analysis, they have also intensified the difficulty of distinguishing between real and AI-produced language. Existing approaches, including traditional machine learning and deep learning models, face limitations in efficiency, scalability, and contextual understanding. Classical ML algorithms are lightweight but lack semantic depth, whereas deep learning and transformer-based architectures capture rich linguistic features at the expense of high computational cost and long training times. This research leverages a hybrid pre-trained Transformer-GRU model to address these limitations. The model uses a pre-trained Transformer solely for feature extraction, capturing deep contextual embeddings without additional fine-tuning. These extracted representations are then fed into a Gated Recurrent Unit network that performs temporal sequence modelling and final classification. By freezing the transformer layers and training only the GRU, the hybrid design combines transformer-based semantic strength with computational efficiency. This structure enables faster convergence, lower resource consumption, and adaptability across multilingual and domain-specific applications. Experimental results show that the DistilBERT + GRU model outperformed baseline ML and DL models, achieving the highest accuracy of 86.12%. The framework aims to deliver an efficient, scalable, and context-aware system suitable for detecting AI-generated text and performing advanced sentiment analysis in real-world NLP environments.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Milosavljevic, Vladimir UNSPECIFIED
Subjects:	Q Science > QH Natural history > QH301 Biology > Methods of research. Technique. Experimental biology > Data processing. Bioinformatics > Artificial intelligence Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Artificial intelligence P Language and Literature > P Philology. Linguistics > Computational linguistics. Natural language processing H Social Sciences > HF Commerce > Electronic Commerce Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4150 Computer Network Resources > The Internet > World Wide Web > Websites > Online social networks T Technology > TK Electrical engineering. Electronics. Nuclear engineering > Telecommunications > The Internet > World Wide Web > Websites > Online social networks
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	07 Sep 2026 09:31
Last Modified:	07 Sep 2026 09:31
URI:	https://norma.ncirl.ie/id/eprint/9850

Actions (login required)

View Item