Skip to content

README.md

Latest commit

 

History

History
156 lines (104 loc) · 8.95 KB

README.md

File metadata and controls

156 lines (104 loc) · 8.95 KB
Raw

SEP Essentiality Check — AI-Assisted Analysis | SEP 必要性檢核 — AI 輔助分析

AI-generated Standard Essential Patent essentiality analysis outputs for methodology discussion and tool development exchange.

AI 生成之標準必要專利必要性分析成果，用於方法論討論與工具開發交流。

⚠️ Legal Disclaimer | 法律免責聲明

All analysis outputs in this repository are generated by AI (Large Language Models). They do NOT constitute legal opinions, legal advice, or any form of professional counsel.

本倉庫中所有分析成果均由 AI（大型語言模型）生成。其不構成法律意見、法律建議或任何形式之專業諮詢。

  • These materials are published solely for the purposes of AI tool development exchange, methodology discussion, and technical research within the IP professional community.

  • No attorney-client relationship, consultant-client relationship, or any professional engagement is created by viewing or using these materials.

  • The AI-generated analyses may contain errors, hallucinations, or misinterpretations. They have NOT been verified to the standard required for litigation, licensing negotiations, or any legal proceeding.

  • Any party relying on these materials for actual legal or business decisions does so entirely at their own risk.

  • 本資料之發布目的僅限於智慧財產權專業社群內的 AI 工具開發交流、方法論討論與技術研究。

  • 瀏覽或使用本資料不會建立任何律師—當事人關係、顧問—客戶關係或任何專業委任關係。

  • AI 生成之分析可能包含錯誤、幻覺（hallucination）或誤解。其未經達到訴訟、授權談判或任何法律程序所要求之標準的驗證。

  • 任何基於本資料進行實際法律或商業決策之當事方，須完全自行承擔風險。

What is This? | 這是什麼？

This repository contains AI-generated analysis outputs from an SEP (Standard Essential Patent) essentiality checking workflow. The prompt/system instructions that produce these outputs are not included in this repository.

本倉庫包含 SEP（標準必要專利）必要性檢核工作流程產出的 AI 生成分析成果。產出這些成果的提示詞 / 系統指令不包含於本倉庫中。

The purpose is to share and discuss:

分享與討論的目的：

  • How AI performs on structured patent-vs-standard mapping tasks

  • Where AI analysis is useful as a preliminary screening tool, and where it falls short

  • Methodological considerations for integrating LLMs into patent engineering workflows

  • The evolving capability boundaries of AI in technical-legal analysis

  • AI 在結構化的「專利 vs. 標準」比對任務中的表現

  • AI 分析在哪些環節可作為初步篩選工具、在哪些環節力有未逮

  • 將 LLM 整合進專利工程工作流程的方法論考量

  • AI 在技術—法律分析中的能力邊界演進

Analysis Output Structure | 分析成果結構

Each published analysis follows a structured format:

每份公開的分析遵循結構化格式：

  1. Executive Summary — Bottom-line essentiality verdict (ESSENTIAL / LIKELY ESSENTIAL / PARTIALLY ESSENTIAL / NOT ESSENTIAL) with confidence level

  2. Claim Chart — Element-by-element mapping between patent claim language and technical standard references, with mapping status for each element

  3. Dialectical Analysis — Thesis (direct mapping conclusion) → Antithesis (counterarguments from patent holder and implementer perspectives) → Synthesis (pragmatic verdict with known unknowns)

  4. Strategic Implications — Cui Bono analysis and actionable next steps

  5. 執行摘要 — 必要性底線結論（ESSENTIAL / LIKELY ESSENTIAL / PARTIALLY ESSENTIAL / NOT ESSENTIAL），附信心水準

  6. Claim Chart — 專利請求項文字與技術標準引用之逐 element 比對，每個 element 標示比對狀態

  7. 辯證分析 — 正題（直接比對結論）→ 反題（專利權人與實施者角度之反駁）→ 合題（含已知未知之務實結論）

  8. 戰略意涵 — Cui Bono 分析與可執行的下一步

Scope & Methodology Context | 範圍與方法論脈絡

The underlying analysis methodology draws from established patent law and SEP practice, including:

底層分析方法論援引既有的專利法與 SEP 實務，包括：

  • Claim construction under the Phillips v. AWH Corp. (2005) standard, with intrinsic evidence hierarchy (claim language → other claims → specification → prosecution history)

  • Element-by-element mapping against normative (mandatory) provisions of technical standards, distinguishing "shall" from "may"

  • Cross-layer analysis — distinguishing between a standard defining a mechanism vs. a standard mandating a specific use of that mechanism

  • Normative vs. informative text differentiation in standards documents (ETSI, IEEE, 3GPP)

  • 依 Phillips v. AWH Corp. (2005) 標準之 Claim construction，遵循 intrinsic evidence 優先順序（請求項文字 → 其他請求項 → 說明書 → 審查歷史）

  • 針對技術標準之規範性（強制性）條文的逐 element 比對，區分 "shall" 與 "may"

  • 跨層分析 — 區分「標準定義了某機制」與「標準要求以特定方式使用該機制」

  • 標準文件（ETSI、IEEE、3GPP）中規範性 vs. 資訊性文本之區分

Repository Structure | 倉庫結構

SEP-essentiality-check-Prompt/
├── README.md
├── reports/            — De-identified analysis reports (v2.0+)
│                         去識別化分析報告（v2.0+）
├── case_studies/       — Judgment-based analysis summaries (planned)
│                         基於判決之分析摘要（規劃中）
└── archive/            — Earlier format outputs (pre-v2.0, historical)
                          早期格式產出（v2.0 以前，歷史參考）

What is NOT in This Repository | 本倉庫不包含的內容

  • ❌ The prompt / system instruction used to generate analyses

  • ❌ Any confidential case information, client data, or privileged materials

  • ❌ Legal opinions or recommendations for specific disputes

  • ❌ Validated or litigation-ready claim charts

  • ❌ 用於生成分析的提示詞 / 系統指令

  • ❌ 任何機密案件資訊、客戶資料或特權資料

  • ❌ 針對特定爭議之法律意見或建議

  • ❌ 經驗證的或可用於訴訟之 claim chart

Intended Audience | 預設讀者

  • Patent engineers and IP professionals exploring AI-augmented workflows

  • Researchers studying LLM performance on technical-legal reasoning tasks

  • SEP licensing practitioners interested in AI as a preliminary screening tool

  • 探索 AI 輔助工作流程的專利工程師與智財專業人士

  • 研究 LLM 在技術—法律推理任務上表現的研究者

  • 對 AI 作為初步篩選工具有興趣的 SEP 授權從業者

Version History | 版本歷史

Version Date Description
v0.1 2025-07 Initial single-pass SEP analysis prompt (included in repo as historical reference)
v2.0+ 2026 Significantly evolved architecture with structured claim construction protocol, multi-tier mapping system, dialectical analysis, and quality control checklist. Outputs from this version are published here; the prompt itself is not public.
v2.4 2026-03 Current version. Added Decision Tree verdict (replacing numerical scoring), programmatic search protocol for large standards (>100p), output de-identification, and dual deployment mode (Claude / general LLM).
版本 日期 說明
v0.1 2025-07 初始單次 SEP 分析提示詞（作為歷史參考收錄於 repo 中）
v2.0+ 2026 大幅演進之架構，包含結構化 claim construction 協議、多層級比對系統、辯證分析與品質管控檢核表。本版產出之分析成果公開於此；提示詞本身不公開。
v2.4 2026-03 現行版本。新增 Decision Tree 判定（取代數字評分）、大型標準文件程式性搜索協議（>100 頁）、輸出去識別化、雙部署模式（Claude / 通用 LLM）。

License | 授權

MIT License

The MIT License applies to the repository structure, README, and any code or scripts. The AI-generated analysis outputs are provided "as-is" under the same license with the additional disclaimer stated above.

MIT 授權適用於倉庫結構、README 及任何程式碼或腳本。AI 生成之分析成果以相同授權「按現狀」提供，並附加上述免責聲明。

Author | 作者

CamusTsai — IP professional with 15+ years of experience in patent litigation and SEP/FRAND licensing. Exploring AI-augmented patent analysis workflows with structured LLM cognitive architectures.

CamusTsai — 擁有 15 年以上專利訴訟與 SEP/FRAND 授權經驗之智慧財產權專業人士。以結構化 LLM 認知架構探索 AI 輔助專利分析工作流程。