FCC Restoring Internet Freedom Proceeding (Net Neutrality Deregulation)
Historic debate on classification of broadband internet providers as common carriers under Title II of the Communications Act.
Semantic Clustering vs. Exact-Hash Baseline
Organized campaigns bypass exact duplicate filters by randomly changing paragraph prefaces, using synonym matrices, or injecting individual grievances.
By generating dense embeddings and running connected-components over cosine similarities, Astroturf captures mutated comment versions. In this landmark FCC rulemaking, naive string matching surfaced only 16 comments, while semantic clustering mapped 1,002 comments - representing a 62x detection lift.
Filing Velocity Spike Analysis
A classic symptom of automated astroturfing is the temporal spike. Real grassroots citizens file comments smoothly across days, weeks, and months. Bot campaigns fire bulk API imports or schedule macro pipelines, creating vertical volume walls.
Hourly analysis reveals that 94.2% of the campaign comments on August 28, 2017, were filed in a single dense burst window, signifying automated machine deployment.
Discovered Campaigns
Showing 3 campaignsMethodology & Ingestion Disclosures
Data Sourcing
Comments are fetched directly from official federal APIs (regulations.gov v4 API or the FCC ECFS public portal). Raw comments are ingested into our Delta Lake Bronze schema, stripping HTML noise and validating submitter fields without discarding critical metadata.
Clustering Limits & Caveats
Semantic clustering groups comments based on high-dimensional text vectors. It uses a similarity threshold (0.92 cosine proximity) to guarantee that only text sharing core structural, political, and semantic boilerplate is linked. False positives are audited against strict quality benchmarks.