org_committee_links

One row per (commenter organization name, FEC committee) name match, derived by build_org_committee_links. Materializes the organization-name bridge between the regulations.gov corpus and fec_committees — the join the data model always described but left to each query author, so every consumer normalized names differently. organization is the raw string as filed, joining straight back to comments.organization; committee_id joins to fec_committees. Coverage is inherently small: comments.organization is populated on only ~0.08% of comments (~20.7K of ~25.8M), and most commenting organizations do not run a federal PAC, so a few hundred organizations resolving is the correct answer rather than a matcher to tune harder. Matching runs in three tiers (exact, core, prefix) and every row carries match_method, confidence, and committee_match_count so consumers pick their own precision bar instead of trusting an opaque score. A high committee_match_count is usually a real affiliate network (Planned Parenthood matches ~90 state committees), not an error — which is why such rows are labelled rather than dropped.

• Parquet file: org_committee_links.parquet
• Queryable via MCP query_sql: Yes

Column	Type	Description
organization	VARCHAR	Commenter organization exactly as filed on the comment. Join key back to comments.organization.
organization_norm	VARCHAR	organization uppercased with parenthetical asides, apostrophes and punctuation removed and & expanded to AND.
organization_core	VARCHAR	organization_norm with trailing legal suffixes (INC, LLC, ...) and a leading THE removed. The form the core/prefix tiers compare.
name_source	VARCHAR	Where the organization name came from. Always organization_field today; text-derived names (comment title, letterhead, signature block) would be added as extra rows under their own source.
committee_id	VARCHAR	Matched OpenFEC committee identifier. Joins to fec_committees.committee_id.
committee_name	VARCHAR	Matched committee name as registered with the FEC.
committee_type_full	VARCHAR	Human-readable committee type of the matched committee (e.g. PAC - Qualified).
designation_full	VARCHAR	Human-readable committee designation of the matched committee.
party_full	VARCHAR	Human-readable political party of the matched committee. Often null for non-party committees.
organization_type_full	VARCHAR	Sponsoring organization type of the matched committee (e.g. Trade Association, Labor Organization). Often null.
committee_state	VARCHAR	Two-letter state on the matched committee (fec_committees.state). Often null.
match_method	VARCHAR	How the pair matched: exact (full normalized names equal), core (decoration-stripped cores equal), or prefix (committee core starts with the whole organization core on a token boundary).
confidence	VARCHAR	high for exact/core; for prefix, medium when committee_match_count <= 5 and low above that. Filter on this to pick a precision bar.
committee_match_count	BIGINT	How many committees this organization matched in total. High values are usually genuine affiliate networks (unions, Planned Parenthood); treat with confidence.
comment_count	BIGINT	Comments filed under this exact organization string (deduplicated on comment_id, newest modify_date wins — matching the MCP comments view).
docket_count	BIGINT	Distinct dockets this organization string commented on.
agency_codes_json	VARCHAR	JSON array of the distinct agency codes this organization commented to, sorted.
first_comment_date	VARCHAR	Earliest posted_date across this organization's comments (ISO 8601 string).
last_comment_date	VARCHAR	Latest posted_date across this organization's comments (ISO 8601 string).