EUVIMEDEUVIMED entdecken
Crossref

CUTTING THE RIGHT WIRE: A Large-Scale Machine Learning Approach to Detect Review Bombing on Steam

Marcus V. G. Ribeiro, Philipe F. Melo, Julio C. S. Reis

Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment · 08.11.2026 · 22 (1) · 170-180

DOI: 10.1609/aiide.v22i1.43575 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, online erschienen am 08.11.2026. Verfasst von Marcus V. G. Ribeiro und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Steam is the largest digital distribution platform for PC games and hosts one of the largest collections of user game reviews. Within this ecosystem, Review Bombing (RB) has emerged as a coordinated behavior in which large volumes of reviews are posted over short periods to influence a game's public evaluation. Despite growing attention from researchers and industry, automatically detecting RB episodes remains challenging due to their temporal, behavioral, and semantic characteristics. In this work, we propose a machine learning approach for detecting RB episodes on Steam. Our study is based on a dataset comprising 157,825 games and more than 103 million user reviews, including 89 RB episodes and 11,800,720 reviews from RB episodes identified by Steam's official moderation pipeline and manually validated. We evaluate multiple machine learning models using temporal, semantic, and behavioral features extracted from review activity. The results demonstrate both the feasibility and the challenges of automatically detecting RB, achieving an F1 score of 0.67 with traditional machine learning classifiers. Feature importance analysis shows that semantic shifts in review content, temporal variations in reviewing activity, and the concentration of negative reviews are the most relevant predictors of RB incidents.

Zitieren

APA

Ribeiro, M. V. G., Melo, P. F., & Reis, J. C. S. (2026). CUTTING THE RIGHT WIRE: A Large-Scale Machine Learning Approach to Detect Review Bombing on Steam. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment, 22(1), 170-180. https://doi.org/10.1609/aiide.v22i1.43575

Vancouver

Ribeiro MVG, Melo PF, Reis JCS. CUTTING THE RIGHT WIRE: A Large-Scale Machine Learning Approach to Detect Review Bombing on Steam. Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment. 2026;22(1):170-180. doi:10.1609/aiide.v22i1.43575

BibTeX
@article{Ribeiro2026,
  author = {Ribeiro, Marcus V. G. and Melo, Philipe F. and Reis, Julio C. S.},
  title = {CUTTING THE RIGHT WIRE: A Large-Scale Machine Learning Approach to Detect Review Bombing on Steam},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence and Interactive Digital Entertainment},
  year = {2026},
  volume = {22},
  number = {1},
  pages = {170--180},
  doi = {10.1609/aiide.v22i1.43575}
}
Kostenlos starten →