當證據跨越學籍：SDG 4、台灣高教與綠領就業的連續性

EIA-017 原稿於 2026 年 9 月 30 日發表，提出一個超越學籍和雇用關係的問題：人在大學修課、研究、參與社區或永續活動，畢業、離職或轉換機構之後，相關紀錄能否仍保留來源、時間、發證者與活動脈絡？原稿稱之為機構證據記憶、個人證據連續性、參與證據連續性及終身證據連續性；這些是 EMJ.LIFE 的分析概念，不是 UNESCO、歐盟或其他標準組織採納的官方名詞。SDG 4.7 關注永續發展知識與技能，UNESCO 的全機構方法則把教學、研究、治理、營運和社區連結放在同一教育環境。SNN.TW 的台灣解讀要問：校務活動形成的證據如何在畢業後仍可被理解，又如何避免把參與紀錄直接宣稱為技能、認證或就業成效？

The original EIA-017 was published on 30 September 2026. It asks whether evidence made during study, research, community work or employment can keep its origin, time, issuer and activity context after graduation or a change of employer. Its four terms—Institutional Evidence Memory, Individual Evidence Continuity, Participation Evidence Continuity and Lifelong Evidence Continuity—are EMJ.LIFE analytical propositions, not vocabulary adopted by UNESCO, the EU or a technical standards body. SDG 4.7 concerns knowledge and skills for sustainable development. UNESCO's whole-institution approach places teaching, research, governance, operations and community engagement within a wider learning environment. The Taiwan question is how university activity can remain intelligible after an affiliation ends without turning a participation record into an automatic claim of competence, certification or an employment outcome.

制度背景分成三層。第一，UNESCO 於 2026 年 9 月發布永續教育全機構指引，SDG 4.7 將永續知識與技能置於教育目標；大學的活動來源不只在永續辦公室，還分散在課程、研究、校園營運和社區。AASHE STARS 提供校務永續評估的跨領域例子，但它的評分或採認不是源頭活動的替代品。第二，歐洲理事會的微證書建議與 Europass 數位學習證明，及 1EdTech CLR 2.0、W3C 可驗證憑證資料模型，顯示學習成就可攜、跨發證者彙整與機器驗證的技術與治理路徑；加密驗證只驗出資料簽發和完整性，不證明聲稱必然真實。第三，THE 與 QS 的高教永續評比是不同下游用途，各自決定方法與准入。原稿以 QS Sustainability 2026 Top 500 作初步邀約母體界線，不能解讀為 QS 背書、任何大學採用或評比分數改善。 這些架構的權威不可混用：教育目標說明應培育什麼，建議與技術標準說明可如何記錄或交換，校方治理程序決定哪一種參與可被確認，雇主再決定該證據與職務的關聯。跨境格式相容不能跳過台灣學校的個資告知、資料保留與學生異議程序，也不會讓評比機構接受未經其方法審查的資料。

There are three distinct institutional layers. UNESCO's September 2026 guidance advances a whole-institution approach to Education for Sustainable Development, while SDG 4.7 supplies the educational purpose. Campus sustainability activity can arise in courses, laboratories, operations and communities rather than only in one reporting office. AASHE STARS illustrates a broad destination for institutional assessment, but its scoring is not a substitute for the originating event. Europe's Council Recommendation on micro-credentials, Europass Digital Credentials for Learning, 1EdTech CLR 2.0 and W3C Verifiable Credentials Data Model 2.0 show routes for portable achievements, multi-issuer records and machine verification. A signature can support integrity and issuer verification without proving that every substantive claim is true. THE and QS assess higher education through their own methods and authority. The original uses the published QS Sustainability 2026 Top 500 only as a boundary for initial outreach. That boundary is not QS endorsement, university adoption or a predicted ranking gain. The authorities in this chain must remain separate. Education goals describe what learning should address; recommendations and technical standards describe ways to represent or exchange evidence; the university decides what participation it can attest; and an employer judges relevance to a particular role. Compatibility with a cross-border data format cannot waive a Taiwan institution’s notice, retention, correction and objection process. It also cannot compel a ranking provider to accept a record outside that provider’s methodology.

台灣的結構摩擦出現在兩種時間尺度。校方持續存在，學生、研究助理、社區參與者和職員卻會離校；活動照片、報名表、課程平台、實驗紀錄、社區回饋與單張證書又分屬不同系統。一位學生若以校園信箱登入活動，畢業停用帳號後只剩沒有主辦者、日期與核驗狀態的 PDF，其紀錄或許仍在，卻難以用於合理的後續查核。另一方面，把同一活動灌入學校成果報告與履歷，也可能把『曾參與』膨脹為『具備某項職能』。環境部的綠領培訓與測驗是台灣已有的正式課程與合格證明案例，必須和一般永續活動紀錄區別；政府核發的合格證明有其課程與測驗條件，不能移植到 EDU SDGS PASS。雇主使用職涯資料時也應區分出席、完成、評量、貢獻及核發資格，避免以單一徽章作錄用結論。 若活動由校外單位共同主辦，校方還須辨認誰可更正出席名單、誰有權證明實際任務、誰能接收個人資料。當撤銷或更正未同步到個人持有的副本，舊紀錄雖然能通過簽章檢查，仍可能誤導下一位查核者。證據連續性因此包含後續狀態的治理，不只是讓檔案永遠可下載。

Taiwan faces a mismatch of time horizons. A university persists while students, assistants, researchers, employees and community participants move on. Course platforms, event registrations, photographs, laboratory records and community feedback may sit in separate systems. When a student loses access to a campus account, a remaining PDF without the organiser, date or verification status may have little evidential value. Conversely, copying one event into a university report and a CV can inflate attendance into a skill claim. Taiwan's Ministry of Environment offers a useful boundary case: its green-skills course has defined training hours and a test for an official completion credential. That formal route is separate from ordinary sustainability participation and cannot be conferred on EDU SDGS PASS. An employer should distinguish attendance, completion, assessment, contribution and issuance eligibility, then decide the relevance to a particular role. A single badge cannot make the hiring decision. When an external organisation co-hosts an activity, the institution must identify who can correct the attendance list, who can attest the participant’s actual task and who may receive personal data. If a correction or revocation does not reach a learner-held copy, the old record might still pass a signature check while misleading the next reviewer. Continuity therefore requires governance of later status, not merely permanent file availability.

可行的設計先以一次活動作最小單元，建立雙向但權限分明的關係：校方保存活動和機構脈絡，個人可持有經授權的參與紀錄。原始層保存課程或活動目的、主辦單位、來源檔與取得時間；身分層保存機構識別、當時學籍或角色及發證者；事件層保存日期、地點、參與方式、完成條件、評量方法和版本；控制層保存同意、可見範圍、撤銷與更正紀錄；使用層則為每個校務揭露、認證或雇用情境分別建立採認決定。永久保存不等於永久有效，個人轉校、畢業或轉職後，查核者仍須看當時狀態和後來的變更。可以先在一門 SDG 4.7 相關課程或社區專案試行，再測試學生下載紀錄、校方更正事件、受邀雇主核對來源，並量測取回率、錯誤更正與權限撤回。這是 SNN.TW 編輯提出的驗證設計，不是已部署的共同技術規格。

A practical design starts with one activity as the smallest auditable unit and maintains two governed relationships. The institution retains the event and its institutional context; the person can retain an authorised record of participation. An origin layer keeps the course or project purpose, organiser, primary file and capture time. An identity layer keeps the institution identifier, historical affiliation, role and issuer. An event layer keeps date, location, mode of participation, completion conditions, assessment method and version. A control layer keeps consent, access scope, revocation and correction history. A use layer records a separate admissibility decision for institutional reporting, accreditation, employment or another destination. Persistent retention does not mean permanent validity. After a learner changes school or employment, a verifier needs both the historical status and subsequent changes. A pilot could use one SDG 4.7 course or community project, testing a learner's retrieval, an institutional correction and a consented employer verification. Measure retrieval success, errors and revocation. This is an SNN.TW editorial validation design, not a jointly issued technical specification or evidence of deployment.

證據鏈分成獨立外部證據與受控實作資料。前者以 UNESCO、AASHE、歐盟理事會、Europass、1EdTech、W3C、THE、QS 及台灣環境部的具日期原始頁面支持教育目標、全機構活動、可攜證明、評估用途與台灣培訓和就業脈絡；15 筆有標題、機構、網址、日期、查核時間與有限支撐範圍的外部紀錄附在 CMS。聯合國 SDG 4 目標頁與原稿的 WHED 識別說明可作背景連結，但未以無法確認的頁面發布日充作 15 筆門檻。後者是.net EIA-017、EMJ.NEXUS 與 SDGS PASS 公開架構，只能支持 EMJ.LIFE 的概念與 2026 年 9 月的測試範圍。9 月 28 日 Student Personal 路徑開始運作，與數所台灣大學的邀約接觸仍屬尚待校方確認的溝通，不能推論受邀者同意、採用、導入、核准或已完成任何正式資源移轉。這項面向全球高等教育的 SDG 4 公共利益、非營利取向倡議，應與既有環境部收費培訓、正式證書及任何法律地位分清。 對 15 筆外部來源的使用也應逐項限縮：技術規範支持資料形式，教育機構資料支持活動場景，政府就業調查支持台灣市場背景；三者不能互相替代。環境部課程的參訓及合格人數不應用來估計 EDU SDGS PASS 的轉換率。受邀學校的回覆、導入與學生同意須由後續第一手紀錄獨立核對。

The chain separates independent external evidence from controlled implementation records. Fifteen dated official records from UNESCO, AASHE, the EU Council, Europass, 1EdTech, W3C, THE, QS and Taiwan's Ministry of Environment support the bounded claims about educational purpose, institutional activity, portable credentials, assessment destinations and the local training and labour market. Every counted CMS record has a title, institution, URL, publication date, access time and claim scope. The UN SDG 4 goal page and the original's WHED identity explanation provide contextual links, but an undated web page is not assigned a fabricated publication day to meet the threshold. The.net EIA-017 and EMJ.NEXUS/SDGS PASS pages are first-party context for EMJ.LIFE's concepts and September 2026 test environment, not independent effectiveness evidence. The Student Personal route began on 28 September. Invitations to several Taiwan universities remain outreach pending each institution's own response; they do not show acceptance, deployment, approval, adoption or a completed transfer. The SDG 4 initiative has a stated non-profit public-interest orientation for global higher education. It must not be conflated with a legal tax status, a completed transfer, the separate fee-bearing ministry course or an official credential. The fifteen external records have distinct claim boundaries. Technical specifications support the data form, institutional education sources support the activity setting, and official Taiwan employment releases support labour-market context. None can stand in for the others. Enrolment and examination results from a Ministry of Environment course cannot be used to estimate conversion for EDU SDGS PASS. Responses by invited universities, implementation and student consent would require later first-hand records.

教育市場的第一個可能影響是校務資料治理。台灣大學的 USR、永續課程、研究與校園活動可能面對多種報告及評比請求，同一事件的來源可以重用，評分或揭露判斷仍由各目的地決定。第二是個人學習歷程與就業銜接。環境部 2026 年上半年綠領職缺調查呈現對綠色技能的市場需求，部會與大學合作的 48 小時課程和測驗則是有條件的正式證明實例；這些數據不證明一般永續參與紀錄會增加任何人的錄取率或薪資。第三是雇主盡職查核。製造、營建、金融與服務企業可在徵才或內訓中核對發證者、活動、能力評量和有效狀態，但應經個人授權並限定用途。跨境的大學聯盟、歐洲憑證系統與全球評比可能加深對可攜證據的期待；這只是機制推論，須以台灣學校制度、實際雇主採認規則與學生同意紀錄驗證。

The first possible Taiwan effect is university evidence governance. USR activity, sustainability courses, research and campus operations may be requested by several reporting and assessment destinations. The event source can be reused; each methodology still makes its own decision. The second is education-to-work continuity. Taiwan's Ministry of Environment reported green-skills job demand in its 2026 first-half survey; its university-linked 48-hour training and examination are an example of a credential with explicit conditions. Those figures do not show that an ordinary activity record increases anyone's hiring probability or pay. The third is employer diligence. Manufacturers, construction firms, finance companies and service employers may verify issuer, activity, actual assessment and current status for recruitment or training, subject to the individual's consent and purpose limits. Cross-border learning, European credential practices and global university assessments may increase interest in portable evidence. That is a mechanism-based inference, to be tested against Taiwan university rules, actual employer acceptance criteria and student consent records.

SNN editorial interpretation：全球制度變動的事實是 SDG 4.7 與 UNESCO 全機構永續教育的方向、歐洲微證書及數位學習證明的可攜架構，並非對台灣大學新增單一強制申報義務。傳導機制是跨境學習與求職、校務永續評估、企業綠領徵才及對憑證來源的查核，把持續可理解的活動資料帶入台灣教育和就業市場。台灣的大學教務、USR 與資訊單位可能須協調課程、社區實作和個人同意；學生畢業後需要可取回且可更正的歷程；雇主只能在取得適當授權後判斷特定職務要求。這些是台灣市場影響的待驗推論，不能把 9/28 邀約列為採用成效。揭露前證據基礎設施（Pre-Disclosure Evidence Infrastructure）的最低要求是：在下一次課程或專案結案前，校方責任人應保存活動原檔、時間、主辦者、參與角色、評量條件、權限和版本。在學生畢業或帳號停用前，教務與資訊單位應測試取回、驗證、更正和撤銷紀錄。在雇主引用紀錄前，人資及職務主管應核對發證者、具體能力證據、同意範圍與時效，拒絕把單純出席當成技能認證。若校務報告、雇用或評比的用途改變，需重新判斷可用性。這是 SNN.TW 的制度設計建議，不是 UNESCO、歐盟或評比機構的指令。

SNN editorial interpretation: the global institutional change is the SDG 4.7 and UNESCO whole-institution education direction together with Europe's micro-credential and digital learning credential architecture. These developments create no single new mandatory Taiwan university filing rule. The transmission mechanism flows through cross-border learning and recruitment, university sustainability assessment, green-skills hiring and employer verification. They could carry an expectation of interpretable, persistent activity records into Taiwan's education and employment markets. University registrars, USR teams and IT owners may need to reconcile courses, community projects and consent; graduates need retrievable and correctable histories; employers can judge specific job requirements only with appropriate authority. These are testable Taiwan market implications, not evidence that the 28 September invitations produced adoption. The Pre-Disclosure Evidence Infrastructure requirement is concrete. Before the next course or project closes, the institutional owner should preserve primary files, event time, organiser, participant role, assessment conditions, permission and version. Before graduation or account expiry, registrar and IT teams should test retrieval, verification, correction and revocation records. Before citing a record in hiring, an HR manager and the role owner should verify issuer, actual competence evidence, consent scope and timeliness, declining to treat attendance as certification. A new reporting, employment or ranking purpose triggers a fresh admissibility decision. This is SNN.TW editorial guidance, not a directive from UNESCO, the EU or any ranking body.

下一步應觀察可驗證的連續性，而非宣稱倡議已帶來效果。試點可設定同一活動的機構與個人紀錄能否在學生離校半年後由授權者重建：原始來源、發證與活動日期、角色、完成條件、同意及更正是否完整；再由未參與編製者測試跨校轉換與雇主查核，記錄缺件率、撤銷時間和錯誤申訴。學校是否加入、參與者是否持續使用、雇主是否採認、就業是否改善，都需分開的觀察資料及適當對照，不能從 2026 年 9 月的架構頁或邀約推導。SDG 4 是此倡議的教育目的軸線；受邀大學不必逐一公開點名，任何後續合作、費用、權利或正式資源移轉狀態都應待文件及權責流程完成後另行核實。 評估時還應分組記錄學校可保留的機構證據與個人可攜帶的紀錄，避免把沒有同意分享的學生列為遺失資料。查核人應能區分來源不存在、來源存在但沒有權限、來源已更正、以及活動本身未達完成條件。只有當這些失敗型態可被辨別，缺件率與重建時間才有比較意義。

The next observation should test continuity instead of claiming impact. A pilot can ask whether an authorised reviewer can reconstruct the same institutional and individual event six months after departure: primary source, issuer and event dates, role, completion condition, consent and corrections. Someone outside the preparation team should then test transfer between institutions and an employer's authorised check, logging missing records, time to revoke and complaint resolution. University participation, learner use, employer recognition and employment outcomes need separate observations and suitable comparison; none follows from a September 2026 architecture page or invitation. SDG 4 is the initiative's educational purpose. There is no need to name individual invited universities. Any later partnership, price, rights or formal transfer status should be verified against completed documents and the relevant institutional approvals before publication. Evaluation should separately record evidence retained by the institution and records carried by an individual. A learner who has not consented to sharing should not be counted as missing evidence. Reviewers should distinguish an absent source, an existing source without permission, a corrected source and an activity that never met completion conditions. Only when these failure modes are separable do missing-record rates and reconstruction time become meaningful measures across cohorts.