Capture what is public.
Use the investigator on public Amazon pages. You choose when capture starts and pauses.
See a contribution example →PLATFORM INTEGRITY · ORIGINAL RESEARCH
Coordination leaves
public traces.
A burst of reviews. The same people, appearing again. Follow the structure of coordinated reviewing through the public evidence it leaves behind.
Created by Gijs Overgoor ↗ · Based on research with Ali Tosyali and other coauthors
REVIEWERPRODUCTDATERATING
A BOUNDED INVESTIGATIONSynthetic example
Start with one history.
Follow the overlap.
20histories opened
107reviewers reached
12most repeated overlap
Reviewers reached as the search progresses
Computed by the paper’s Python implementation on generated records. Inspect the evidence →
FROM THE PAPER
92.2%
Reviewers linked to observed campaigns
Reached after 500 histories
12.2%
Explore the archive resultsRecurring patterns in verified five-star reviews
2023 archive · not confirmed fake reviews
and their settings ↗
FROM RESEARCH TO COLLECTIVE EVIDENCE
An investigation others
can build on.
Explore the research, inspect worked examples, and use the investigator to build a case in your browser. Keep a local record of the evidence and its coverage. Shared submissions and reviewed community datasets are planned for a later release.
Follow repeated overlap.
Inspect distinct co-review neighborhoods, product windows, and the reasons behind each count.
Inspect a computed example →Make evidence reusable.
Export your case and ranked reviewers to your own files. Keep incomplete coverage visible as you decide what to examine next.
Get the tool →THE WORK BEHIND THE TOOL
Can public traces help detect coordinated fake review campaigns on digital platforms?
Gijs Overgoor · Ali Tosyali · Ethan Feldman · Anol Bhattacherjee
Read the working paper ↗