Methodology
Built on public data. Validated by science.
Ingestion Pipeline
Ingest
→Resolve
→Graph
→Score
→Verify
Every record flows through entity resolution, graph construction, IES scoring, and XRPL timestamping before publication.
Influence Vortex Model — 8 Dimensions
Financial Contributions
Lobbying Access
Committee Power
Revolving Door
Earmark Allocation
Foreign Interests
Contract Awards
Regulatory Capture
Data Sources — 10 Federal Databases
House PTR DisclosuresSenate eFDFEC RecordsCongressional RecordsLobbying DatabasesFARA FilingsCourt DocumentsDOJ ReleasesIRS 990sUSAID Records
▸
House PTR Disclosures
Periodic Transaction Reports — all stock trades by House members and staff
▸
Senate eFD
Electronic Financial Disclosures for all sitting senators
▸
FEC Records
Federal Election Commission bulk data — all campaign contributions and expenditures
▸
Congressional Records
Official legislative history, committee assignments, and floor votes
▸
Lobbying Databases
LDA filings — all registered lobbyist activity and client relationships
▸
FARA Filings
Foreign Agent Registration Act — foreign influence disclosures
▸
Court Documents
PACER — federal case filings, indictments, and judgments
▸
DOJ Releases
Department of Justice press releases and enforcement actions
▸
IRS 990s
501(c) nonprofit and dark money group filings
▸
USAID Records
Foreign assistance disbursement records and contractor data
Validation & Architecture
IES v3.5 Scoring
Influence Exposure Score integrating donation, contract, and lobbying network proximity.
Validated at AUC 0.9575
XGBoost classifier on 420-dim embeddings (384 semantic + 36 graph-structural). Tested across multiple models.
XRPL Blockchain Provenance
Every ingestion event timestamped on XRPL. Immutable audit trail for all 408K+ entities.
Sentinel Oracle (NLQ Layer)
Fine-tuned Qwen2.5 model for natural language queries against the Neo4j knowledge graph.