DigestAI news desk

Cut through the AI noise.

Research

Survey reviews 211 fake review detection studies from 2018 to 2026

A new survey published on arXiv examines the evolution of fake review detection methods, covering 211 studies released between 2018 and early 2026. The paper analyzes how the field has shifted from traditional machine learning to pre-trained language models (PLMs) and large language models (LLMs). It highlights a dual challenge: LLMs can now generate highly convincing deceptive reviews, yet they…

1 source primary source

Key points

  • Survey covers 211 studies on fake review detection published from 2018 to early 2026.
  • Analyzes shift from traditional ML to PLM and LLM-based detection methods.
  • Identifies open problems including adversarial generation and cross-domain transfer.

The authors organize the literature by evidence source and fusion level, including review text, sentiment, rating behavior, temporal metadata, user-product graphs, and multimodal content. They trace the integration of textual, behavioral, structural, and external knowledge signals. The survey also evaluates performance trends on standard benchmarks such as Amazon, Yelp, and OpSpam, while noting that inconsistent label construction and evaluation protocols limit direct comparability across studies.

The paper concludes by identifying open problems in the field, including adversarial generation, cross-domain transfer, and the need for uncertainty-aware fusion. It emphasizes the growing difficulty of detecting AI-generated deceptive content and calls for more trustworthy evaluation methods to maintain platform integrity.

Read the original at arXiv cs.CL · by Fanji Yang (Guizhou University of Finance and Economics), Huiyao Chen (Harbin Institute of Technology), Xi Yu (Guizhou University of Finance and Economics), Meishan Zhang (Harbin Institute of Technology), Xiaohong Xiao (Guizhou University of Commerce), Mingsen Deng (Guizhou University of Finance and primary sourceOpen source ↗

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Research

All →

Related stories