fss-policy-kg

Tools 10

• fss_policy_kg_check_currency Is a document still in force? Staleness flags plus supersession / revocation links. Returns: A JSON object with doc_id, source_url (the canonical link to the full source document —, cite it so a reader can open the original), is_stale, current_version_status, supersedes,, revoked_by, superseded_by, and basis_changed (present only when a statute this guidance, IMPLEMENTS has moved — a review trigger). An unknown id returns a structured error with, nearest-match suggestions.
• fss_policy_kg_get_concept_mentions Obligations that MENTION a concept, with force and verbatim. Sparse by design. Inexact input (typo/variant) is resolved through the concept fuzzy matcher before giving up; if nothing resolves, nearest concept suggestions are returned. Returns: A JSON envelope; each result has concept, force, verbatim_span (snippet), source_doc,, method, score. `resolved_from` is set when the input was fuzzy-resolved to a concept.
• fss_policy_kg_get_conflicts Adjudicated deontic conflicts between Obligations — pairs where one provision forbids what another permits (class A) or requires (class B). genuine_conflict = opposed and unresolved (rare, the findings); apparent_resolved = opposed but law resolves it (resolution_basis says how). Absence of a conflict is NOT proof of consistency; each verdict is one model's reading. Returns: A JSON envelope; each result has conflict_id, class, verdict, resolution_basis, rationale, (snippet), both obligation ids/docs and their verbatim spans (snippet), and confidence.
• fss_policy_kg_get_corpus_overview The corpus landscape in ONE call — CALL THIS FIRST for any coverage / gap / 'what documents are there' / 'are there gaps in X' question. Returns the document inventory (each with layer, category, issuing agency, obligation_count, and mention-coverage stats), the corpus totals, the served export version, and the layer model — so you can see where obligations concentrate and how well each document is concept-tagged before drilling in. Read the coverage correctly: resolved_mention_edges / concept_reachable_share measure CONCEPT TAGGING, which is sparse by design — a low share means under-tagging, NOT that the document lacks the topic. raw_mention_count vs resolved_mention_edges shows the untagged remainder. Use search_text for authoritative topic presence. Returns: A JSON envelope; results are documents sorted by obligation_count (largest first), each with, doc_id, title (snippet), agency, layer, category, is_stale, obligation_count,, raw_mention_count, resolved_mention_edges, concept_rea
• fss_policy_kg_get_definitions Search the definition layer. Filter by term/verbatim substring and/or source document. Each definition carries its linked Concept and cross-vocabulary grading links. Returns: A JSON envelope; each result has def_id, term_surface, verbatim_text (snippet), source_doc,, concept, graded_matches, defined_by_reference.
• fss_policy_kg_get_document Document metadata, issuing authority, obligation/segment counts, and IMPLEMENTS links (statutes this guidance enacts, or the guidance enacting a statute). Returns: A JSON object with title, source_url (the canonical link to the full source document —, cite it so a reader can open the original), statute_cite, agency, work_id, baseline_date, is_stale,, authorities, segment_count, obligation_count, implements, implemented_by, top_concepts (the, document's most-mentioned concepts via the derived MENTIONED_IN rollup, with counts), and, basis_changed (a review-trigger flag, present only when a statute this guidance implements has, moved). An unknown id returns a structured error with nearest-match suggestions.
• fss_policy_kg_get_lineage Document lineage: the Work it expresses, sibling expressions, the full supersession / revocation chain, and IMPLEMENTS links with the enacted statute's currency. Returns: A JSON object with source_url (the canonical link to the full source document), work_id,, work_title, sibling_expressions, supersedes, superseded_by, revokes, revoked_by, implements,, implemented_by, is_stale, current_version_status. An unknown id returns a structured error, with nearest-match suggestions.
• fss_policy_kg_get_obligations Filter Obligations by normalized force and/or source document. The conflict surface: binding confidentiality duties are force='obligation'/'prohibition' with source_doc='cipsea'. Returns: A JSON envelope; each result has obl_id, force, force_raw, addressee, obligation_object, (snippet), verbatim_span (snippet), source_doc.
• fss_policy_kg_search_concepts Fuzzy-search the concept vocabulary (663 concepts, tiers A/B/C/M) by canonical term or any surface form. Use this to find the right concept before get_concept_mentions, or to check whether the corpus even has a concept. Handles typos, casing, and plural/inflected forms. Returns: A JSON envelope of results, best match first; each has canonical_term, concept_key, tier,, surface_forms, score (0-100), is_suggestion (true when below the match threshold), an, optional `name_status` (present only when the concept's canonical name has a live secondary, title or a pending statutory rename — read it before quoting the name back, because the, surface form you matched may not be the statutory one), TWO, cross-layer counts: mention_count (how many obligations are TAGGED with this concept —, concept-layer coverage, sparse by design) and text_layer_hits (how many text nodes actually, contain the term — text-layer presence); and `related`: up to 5 concepts that co-occur in the, same obligations (deriv
• fss_policy_kg_search_text Lexical case-insensitive substring search over Obligation, Definition, and Segment text. The best first call to locate where a policy phrase appears verbatim across the corpus. Returns: A JSON envelope; each result has label, id, text (snippet), source_doc, force (Obligations).