Citation Parsing and Graphs
Citation Parsing and Graphs
Citation grammar, tokenization, reporters, statutes, rules, short forms, id and supra, resolution, ambiguity, edges, direction, graph storage, temporal slices, validation, and pinned Hertz inbound records.
Structured Visual
Jurisdiction: US; as of 2026-08-28; not legal advice; Render structure, refuse interpretation, cite, abstain, and hand off.
RENDER STRUCTURE · REFUSE INTERPRETATION · CITE · ABSTAIN · HAND-OFF: render structure, refuse interpretation, cite provenance, abstain when unsupported, and hand off to human review.
Scope and honesty note
Jurisdiction: United States computational-law classroom model; source snapshot 2026-08-28; curriculum as of 2026-08-29. Synthetic inputs, code, labels, measurements and outputs are teaching artifacts, not law, legal advice, authority, eligibility, benefits, tax, court, filing, research, ranking, or outcome determinations. Code encodes selected interpretations and can be incomplete, wrong, outdated, biased, overprecise, underinclusive, or non-isomorphic. The system must cite source and version, expose assumptions and gaps, abstain when unsupported, and hand legal judgment to accountable humans.
See the essential structure first
Start with this deliberately incomplete structure, then use the pinned authorities, worked application, exceptions, and handoff below. This deliberately incomplete preview has 4 nodes; exceptions and legal consequences remain in the sourced prose below.
Jurisdiction: US; as of 2026-08-28; not legal advice; Render structure, refuse interpretation, cite, abstain, and hand off.
RENDER STRUCTURE · REFUSE INTERPRETATION · CITE · ABSTAIN · HAND-OFF: render structure, refuse interpretation, cite provenance, abstain when unsupported, and hand off to human review.
Begin with computational-law doctrine
Citation graphs depend on accurate span detection, parsing, normalization and target resolution. Regex alone is useful for candidates but insufficient for short forms, parallel citations, reporter ambiguity and contextual references. Every edge needs source span, target evidence, parser version and temporal snapshot. An edge means one document cites another; it does not establish approval, holding, precedential force or authority. Graph measurements are derived descriptions and must remain separate from legal hierarchy and validity.
Pinned inbound citation edges
The fixture supplies three bounded source-to-Hertz edges and explicitly disclaims authority-weight claims. Pinned source or measurement: “[{"from": "/f-supp-2d/692/0170-01", "fromDate": "2010-03-09", "fromName": "Watkins v. Omni Life Science, Inc.", "to": "/us/559/0077-01", "type": "cites"}, {"from": "/f-supp-2d/712/0924-01", "fromDate": "2010-05-17", "fromName": "In re Arrowhead Capital Management LLC Class Litigation", "to": "/us/559/0077-01", "type": "cites"}, {"from": "/f-supp-2d/718/0805-01", "fromDate": "2010-03-10", "fromName": "Astra Oil Trading NV v. Petrobras America Inc.", "to": "/us/559/0077-01", "type": "cites"}]” Coordinate: case-citations-hertz-inbound-slice as of 2026-08-28; https://www.neochart.com/catalog/cases/us/volume_559/0077_01/index.html; data via neochart.com, snapshot 2026-08.
Separate node measurement
The exact Hertz importance value illustrates that node measurement is downstream of edge construction and remains descriptive. Pinned source or measurement: “{"case_path": "/us/559/0077-01", "importance": 2.1810471734751303e-06}” Coordinate: Hertz Corp. v. Friend case_importance measurement as of 2026-08-28; https://www.neochart.com/catalog/cases/us/volume_559/0077_01/index.html; data via neochart.com, snapshot 2026-08.
Pin the synthetic computational record
A synthetic citation pipeline starts with the three pinned Hertz inbound source records, stores raw citation spans and tokens, normalizes reporters, resolves targets through aliases and metadata, creates evidence-linked directed edges, retains unresolved candidates, builds inbound and outbound queries, computes descriptive fields and logs gold annotations and regressions.
Work the audited application
Each inbound record is re-parsed and resolves to the Hertz center id; the edge direction stays citing source to cited target. A short-form citation without adequate context remains unresolved. A statute-like citation to an impossible title is quarantined rather than normalized to a plausible title. The Hertz importance value is joined only after edge construction and labeled as-of snapshot, not authority.
Read the populated computational artifact
The citation graph record contains document, raw span, start, end, tokens, citation kind, volume, reporter, page, pin cite, title, section, subsection, year, court, signal, short form, alias, normalized cite, candidate target, context, confidence, resolution, unresolved reason, source node, target node, edge type, direction, evidence, parser version, snapshot, inbound query, outbound query, degree field, measurement, gold annotation, error, quarantine, and reviewer. The artifact contains 16 populated rows.
Jurisdiction: US; as of 2026-08-28; not legal advice; Render structure, refuse interpretation, cite, abstain, and hand off.
RENDER STRUCTURE · REFUSE INTERPRETATION · CITE · ABSTAIN · HAND-OFF: render structure, refuse interpretation, cite provenance, abstain when unsupported, and hand off to human review.
Read the complete record
The complete record keeps sources, stated facts, and questions for review separate. Pinned sources and measurements: Verbatim text or bounded snapshot data. case-citations-hertz-inbound-slice as of 2026-08-28: Pinned inbound citation edges: The fixture supplies three bounded source-to-Hertz edges and explicitly disclaims authority-weight claims.. Hertz Corp. v. Friend case_importance measurement as of 2026-08-28: Separate node measurement: The exact Hertz importance value illustrates that node measurement is downstream of edge construction and remains descriptive.. Synthetic computational record: Classroom inputs and intermediate states. Documents: Three opinions contain citations resolving to Hertz plus statute, rule, parallel, short-form and unresolved strings. Parser: Raw span, tokens, normalized cite, candidate targets, court and date context, resolution confidence, manual decision and edge. Graph: Case and statute nodes, cites edge, cited-by view, source and target dates, snapshot, duplicate and retracted-edge history. Audit trace: Semantics, provenance, execution, evidence, limits and handoff. Detection: Regex candidate, lexer, parser grammar, reporter table, section symbol, title, volume, page, year, court, signal and pin cite. Citation kinds: Case, statute, regulation, rule, constitution, secondary source, docket, slip opinion, neutral cite, parallel cite, id, ibid and supra. Normalization: Unicode, spacing, abbreviation, reporter, section range, subsection, title, canonical string, source span and no destructive guessing. Resolution: Exact key, alias, metadata filter, jurisdiction, temporal availability, candidate set, ambiguity, confidence, human resolution and unresolved. Edges: Source, target, type, direction, source span, pin cite, date, parser version, evidence and validity; citation is not approval or precedential force. Graph queries: Inbound, outbound, neighborhood, path, degree, temporal slice, connected component and co-citation; measurement and authority remain distinct. Validation: Gold spans, target accuracy, broken links, impossible titles, self edge, duplicates, sampled errors, regression and change history.
Narrow summary
Parse and resolve citation spans with evidence, keep edge direction and time explicit, quarantine invalid categories, and never translate graph connection into legal authority.