AI system spots fake reviews by combining text, images and user behavior

· Source: News on Artificial Intelligence and Machine Learning · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics · Depth: Intermediate, quick

Summary

A new AI system, detailed in the *International Journal of Information and Communication Technology* in 2026, detects fake e-commerce reviews by integrating text, images, and reviewer behavior. This multimodal approach addresses the increasing sophistication of deceptive feedback on online marketplaces. The system employs a text convolutional neural network and a pretrained language model for textual analysis, a residual network for image processing, and incorporates reviewer data. It then combines these diverse signals to identify genuine reviews. A Transformer model can further trace the origins and spread of suspicious feedback. Tests on large-scale datasets demonstrated measurable gains over existing detection methods.

Key takeaway

For e-commerce platform operators and fraud detection teams combating sophisticated fake reviews, this multimodal AI system offers superior detection and traceability. You should consider integrating diverse data streams—text, image, and user behavior—into your fraud prevention strategies to enhance accuracy and identify review origins more effectively.

Key insights

Multimodal AI combining text, images, and behavior significantly improves fake e-commerce review detection.

Principles

Method

Utilize text CNN, pretrained language models, residual networks for images, and reviewer data; then apply a Transformer for traceability.

In practice

Topics

Best for: Research Scientist, AI Scientist, Director of AI/ML, General Interest

Related on AIssential

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by News on Artificial Intelligence and Machine Learning.