View Proposal

Proposer
Arulmurugan Ramu
Title
Federated Fake Review Detection Across E-Commerce Platforms
Goal
To design, build and evaluate a federated learning system in which several e-commerce platforms jointly train a fake-review detector without exchanging their review data, and to measure whether the federated model actually outperforms what each platform could train alone.
Description
Fake reviews are a cross-platform problem tackled with single-platform data. The same paid-review operations post on Amazon, Yelp, Trustpilot and regional marketplaces, so the strongest training signal is spread across companies that are commercial competitors and cannot pool customer data for legal and commercial reasons. Federated learning is the standard proposal for this situation
Resources
Q. McAuley et al. Amazon Reviews 2023 dataset — https://amazon-reviews-2023.github.io/ Yelp Open Dataset — https://business.yelp.com/data/resources/open-dataset/ (registration and licence agreement required)
Background
Url
External Link
Difficulty Level
High
Ethical Approval
None
Number Of Students
1
Supervisor
Arulmurugan Ramu
Keywords
federated learning, fake review detection, natural language processing, privacy-preserving machine learning, e-commerce, non-iid data
Degrees
Bachelor of Science in Computational Sciences and Software Engineering
Back to List