code wiki / _hdl_build / nx_research_relations.nx
nx_research_relations.nx
buildroot/runtime/_hdl_build/nx_research_relations.nx
about
nx_research_relations.nx -- Researcher EXTRACT, rung 2: DOCUMENT-level claim
extraction. Splits a document into sentences and runs the rung-1 relation
organ on each, collecting a CLAIM SET (subject -> relation -> object). This is
what turns per-sentence extraction into research depth: ~one qualitative claim
per relational sentence across a whole page, vs the number-only extractor's
handful of digit-facts. Inspectable, sovereign, gated. license_tier: ORIGINAL.
dependencies 4 imports · 3 importers
imports: nx_syscalls.nxnx_research_extract.nxnx_research_relation.nxnx_research_relation_clean.nx
imported by: nx_rel_measure.nxnx_research_landscape.nxnx_research_relations_test.nx
structs
| none |
consts
| 12 | const K_MAGIC_2048: i64 = 2048 |
functions
| 15 | func rel_is_sent_end(c: u8) -> i64 |
| 24 | func rel_app(out: *u8, off: *i64, out_cap: i64, s: *u8) -> i64 |
| 36 | func nx_rel_extract_doc(text: *u8, n: i64, out: *u8, out_cap: i64) -> i64 |
| 73 | func nx_rel_extract_doc_clean(text: *u8, n: i64, out: *u8, out_cap: i64) -> i64 |