ai-cfia / ai-cfia/fertiscan

Redefine label review lifecycle to cover extraction and compliance

未關閉
#680 0 則留言 0 個 reaction 已指派 0 人 在 GitHub 檢視
fullstack
主要語言
HTML
星號
5
分支
2
平均合併
1 天 7 小時
30 天內合併 PR
1

描述

# Context

`ReviewStatus` (`not_started` / `in_progress` / `completed`) was modelled before compliance checks existed. It now conflates extraction, human data review, and compliance evaluation into one opaque field, and the user cannot tell from any given label where it is in the workflow or what action is theirs to take.

# Outcome

An agreed vocabulary and state model for a label's lifecycle that covers extraction, compliance evaluation, and human review, and maps cleanly onto what the user sees.

# Acceptance criteria

- Every stage a label can be in is named and defined
- Every transition between stages is defined, including its trigger (automatic vs. user-driven) and preconditions
- The meaning of "review" is unambiguous once compliance exists
- The model defines what happens to compliance verdicts when extracted data is edited after evaluation
- The model defines what "done" means for a label and what is required to reach it
- The model defines who can advance a label between stages, or explicitly defers that question
- The UI representation is defined: a user opening a label can tell at a glance what stage it is in and what action is expected of them
- Existing terminology inconsistencies between backend enums, API, and UI strings are resolved

貢獻指南

這個儲存庫沒有索引到貢獻指南

評估

這個 Issue 還沒有評估資料。

把新 issue 寄到你的電子郵件信箱

精選適合新手參與的 GitHub issue 摘要。