Burst-Aware Hybrid Temporal–Graph Model for Yelp Fraud Detection

DOI

10.2991/978-94-6239-723-1_15How to use a DOI?

Keywords

fake review detection; Yelp fraud detection; opinion spam; temporal modelling; burst detection; graph neural networks; multi-relation graphs; hybrid learning; class imbalance; robustness

Abstract

Online review platforms are increasingly targeted by coordinated fraud campaigns that distort ratings and mislead consumers. Existing fake-review detectors often rely on either content signals or static behavioural features, which can underperform when spammers camouflage writing style and manipulate interaction patterns. This paper proposes a Burst-Aware Hybrid Temporal–Graph Model for Yelp fraud detection that jointly captures (i) engineered behavioural footprints, (ii) multi-relation collective evidence from review graphs (e.g., shared-user, shared-business/rating, and shared-business time-bucket links), and (iii) temporal burst dynamics that characterize campaign-driven activity. The framework employs an attribute encoder to learn review-level representations, a relation-aware graph encoder to model network effects, and a temporal/burst module that constructs time-ordered contexts and burst statistics within a defined window. A temporal-aware fusion layer integrates these complementary embeddings for robust binary classification under class imbalance using an imbalance-aware loss. Comparative analysis with representative baselines suggests that incorporating burst-aware temporal cues is especially beneficial for improving minority-class capture, thereby strengthening Recall and F1 while maintaining high Precision. The proposed methodology is designed for reproducible evaluation on public Yelp-style benchmarks and supports ablation studies to quantify contributions from temporal, structural, and behavioural components.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Ruchi Jain
AU  - Ajit Kumar Jain
PY  - 2026
DA  - 2026/07/14
TI  - Burst-Aware Hybrid Temporal–Graph Model for Yelp Fraud Detection
BT  - Proceedings of the International Conference on Responsible, Risk-aware, and Regulated AI (RRRAI 2026)
PB  - Atlantis Press
SP  - 163
EP  - 174
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-723-1_15
DO  - 10.2991/978-94-6239-723-1_15
ID  - Jain2026
ER  -