[Fix] NLG Judge 평가 기준을 서비스 정책에 맞게 정렬

[whc9999]

✨ 어떤 이유로 PR를 하셨나요?

• feature 병합
• 버그 수정(아래에 issue #를 남겨주세요)
• 코드 개선
• 코드 수정
• 배포
• 기타(아래에 자세한 내용 기입해주세요)

📋 세부 내용 - 왜 해당 PR이 필요한지 작업 내용을 자세하게 설명해주세요

• MISSED_ANALYSIS 판정 기준을 명확한 문제 문장과 첨삭 가치 손실 중심으로 강화
• 명확한 문제 부재와 MISSED_ANALYSIS가 공존하지 않도록 Judge 프롬프트 보강
• MISSED_MISSING_KEYWORD 평가에서 자격증·면허·학력·경력 연차 등 정형 자격요건 제외
• JD에 없는 키워드와 OR 자격요건의 미선택지를 누락으로 보지 않도록 규칙 추가
• empty questionAnalyses 입력 시 평가 배열 생성 금지 계약 재강조
• overallUsefulness와 noAnalysisAppropriateness 1~5 rubric 구체화
• Judge 프롬프트 contract 테스트 추가

📸 작업 화면 스크린샷

⚠️ PR하기 전에 확인해주세요

• 로컬테스트를 진행하셨나요?
• 머지할 브랜치를 확인하셨나요?
• 관련 label을 선택하셨나요?

🚨 관련 이슈 번호 [#150 ]

Summary by CodeRabbit

변경 사항

• 개선

  • NLG 평가 기준을 더욱 엄격하고 구체적으로 정비했습니다.
  • 분석 누락, 핵심 키워드 누락, 답변 유용성에 대한 평가 범위와 판정 기준을 명확히 했습니다.
  • 입력 분석이 없을 때 불필요한 평가를 생성하지 않도록 출력 규칙을 강화했습니다.
  • 평가 오류 코드와 근거 설명의 일관성을 높였습니다.

• 테스트

  • 주요 평가 규칙과 점수 범위가 프롬프트에 포함되는지 검증하는 테스트를 추가했습니다.

[coderabbitai]

Warning

Review limit reached

@whc9999, you've reached your PR review limit, so we couldn't start this review.

Next review available in: 45 minutes

Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available.
You're only billed for reviews past your plan's rate limits ($0.25/file).

How can I continue?

After more reviews become available, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews.

How do review limits work?

CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability.

For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window.

Please refer docs for additional details.

Review details

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: e9c0af07-9075-467e-ade7-5e78ad483954

📥 Commits

Reviewing files that changed from the base of the PR and between d54e0a8 and 7a269e4.

⛔ Files ignored due to path filters (2)

• evaluation/evaluation_nlg_judge_policy_alignment.csv is excluded by !**/*.csv
• evaluation/evaluation_nlg_judge_policy_alignment_comparison.csv is excluded by !**/*.csv

📒 Files selected for processing (2)

• src/main/java/com/jobdri/jobdri_api/domain/analysis/evaluation/NlgEvaluationAiClient.java
• src/test/java/com/jobdri/jobdri_api/domain/analysis/evaluation/NlgEvaluationAiClientTest.java

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/nlg-judge-policy-alignment

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[coderabbitai]

Caution

Failed to replace (edit) comment. This is likely due to insufficient permissions or the comment being deleted.

Error details

putComment timed out

[coderabbitai]

Actionable comments posted: 2

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In
`@src/main/java/com/jobdri/jobdri_api/domain/analysis/evaluation/NlgEvaluationAiClient.java`:
- Around line 129-159: Refactor buildPrompt by extracting the MISSED_ANALYSIS
and MISSED_MISSING_KEYWORD rule text into separate private static final String
constants or focused helper methods, then compose those elements from
buildPrompt. Preserve the existing rule wording and prompt order while reducing
buildPrompt’s responsibility to assembling the prompt.
- Around line 94-97: Update the missingKeywords evaluation instructions in
NlgEvaluationAiClient so the rule that assigns MISSED_MISSING_KEYWORD when
actual=[] explicitly excludes structured qualification requirements such as
certifications, licenses, education, required experience years, and age. State
that the later exclusion policy takes precedence, preserving
MISSED_MISSING_KEYWORD only for non-structured missing JD/answer requirements.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Pro Plus

Run ID: cb881c4d-c585-482b-a984-710eee5da55e

📥 Commits

Reviewing files that changed from the base of the PR and between a2dad1f and d54e0a8.

⛔ Files ignored due to path filters (2)

• evaluation/evaluation_nlg_judge_two_pass_provenance_strength_fix.csv is excluded by !**/*.csv
• evaluation/evaluation_nlg_judge_two_pass_provenance_strength_fix_comparison.csv is excluded by !**/*.csv

📒 Files selected for processing (2)

• src/main/java/com/jobdri/jobdri_api/domain/analysis/evaluation/NlgEvaluationAiClient.java
• src/test/java/com/jobdri/jobdri_api/domain/analysis/evaluation/NlgEvaluationAiClientTest.java