실수집 코퍼스
실수집 SEC 원문 · 검토 링크.
데이터 읽는 법
Visualization Plan
review-firstLifecycle Frame
one paper, one compliance questionWhy Raw-to-Raw Pairing?
why not clause labels onlySo What?
review action, not just dataContract-to-patent review links
Link contract text directly to patent candidates.
How to Read This
contract → patent → reviewer actionContract-mentioned patents
from exhibit textMSDS review candidates
heuristic linksT1 Pair Candidate Queue
review + hard negativesSEC contract evidence
SEC exhibits and clause candidates
t044 SEC Clauses
clause-levelEDGAR Exhibits
raw docsGrant-back Clause Candidates
annotation-readyAnnotation Sample Pack
balanced triage setAnnotation queue
Label review candidates and export a training CSV.
MSDS patent candidates
Real patent rows from MSDS indexes
Source ledger
Reused local data, gap-fill collection, and remaining risks
Dataset Samples
review before claimPaper3 DART/EDGAR
reusedRisks
annotation gateSubmission Status
external portal gateDraft & Status Docs
writing progressRetrieval Baseline
BM25 / TF-IDF sanity checkExternal submission gate