meta data for this page
  •  

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision		Previous revision
educator:ai-tools [2026/05/18 06:19] – Initial content claudeeducator:ai-tools [2026/06/02 06:46] (current) – Update: AI-01 through AI-06 all live, PHP pipeline, review queue claude
Line 3: Line 3:
 ===== Overview ===== ===== Overview =====
  
-LEAST includes AI-assisted tools that help educators and platform administrators enrich lesson metadata, tag content with roles and skills, and surface discovery connections.+LEAST includes a full AI enrichment suite backed by the Anthropic Claude API. It helps educators and administrators enrich lesson metadata, classify content, surface discovery connections, and govern AI-generated material before it goes live.
  
-===== Lesson Enrichment Pipeline =====+===== Live AI Features =====
  
-The AI pipeline analyses lesson content and suggests: +^ Code ^ Feature ^ Status ^ Where to find it ^ 
-  * Relevant **occupational role tags** (from a taxonomy of 701+ roles) +| AI-01 | Lesson Markup — AI Insights modal | **Live** | Lesson page → AI Insights button | 
-  * **EKOS competency links** (European Key Occupation Standards — 215 competency areas) +| AI-02 | IWTB Role Enrichment — 700+ career roles | **Live** | Admin → IWTB → Role list | 
-  * **Learning objective improvements** +| AI-03 | Role-Lesson Semantic Graph — 734 scored links | **Live** | Admin → IWTB → Role-Lesson links | 
-  * Related lesson suggestions+| AI-04 | Editorial Governance Workspace | **Live** | ''/educator/ai-review-queue.php'' | 
 +| AI-06 | EKOS Classification — UDC + IFQ alignment | **Live** | System → EKOS → Awaiting review |
  
-To run enrichment on a lesson: +===== AI-01: Lesson Markup (AI Insights) =====
-  - Open the lesson → **AI Tools → Enrich** +
-  - Review the suggested tags +
-  - Accept, edit, or reject each suggestion +
-  - Save — accepted tags are applied to the lesson record+
  
-===== Bulk Enrichment =====+Every published lesson has an **AI Insights** button that opens a modal with AI-generated learning support:
  
-For mass tagging of existing lessons: +  * **Overview** — plain-language summary of the lesson 
-  - Navigate to **Educator → AI Enrichment → Batch Run** +  * **Key Points** — bullet list of core concepts 
-  - Select lessons to process (or run on all untagged lessons) +  * **Why This Matters** — real-world relevance 
-  - The pipeline runs in the background; results appear in the **Review Queue**+  * **Suggested Next Steps** — what to explore after this lesson 
 +  * **AI Instructions** — guidance for AI systems using this lesson with learners 
 +  * **Sources** — references to materials used in generation
  
-===== Review Queue =====+Content is reviewed via the AI-04 governance workspace before going live. Stored in ''strLesson_AI_Markup'' and rendered as CommonMark on the lesson page.
  
-AI-generated suggestions always go through a human review step: +===== AI-02: IWTB Role Enrichment =====
-  - Navigate to **Educator → AI Review Queue** +
-  - Each item shows: lesson title, suggested tags, confidence score +
-  - Actions: **Approve**, **Edit and Approve**, **Reject** +
-  - Approved suggestions are applied and logged+
  
-===== Role-Lesson Links =====+The **I Want To Be** (IWTB) engine covers 700+ career roles, each AI-enriched with:
  
-The platform maintains a map of 734 role-lesson connections (AI-generated, human-reviewed). These power: +  * Role background and description 
-  * The **Discover** feature (showing learners content relevant to their role) +  * Suggested next steps for learners 
-  * Employer-facing analytics (what skills does this learning pathway develop?)+  * Key figures in the field 
 +  * Recommended books 
 +  * Related TV shows and films 
 +  * Salary and career outlook data
  
-===== Lesson Content Injection Pipeline =====+Roles are browsable from the learner dashboard and discoverable via the Discover page.
  
-Educators and admins can inject new lesson content at scale using the content pipeline API: +===== AI-03: Role-Lesson Semantic Graph =====
-  - Content descriptors (title, description, URL, type) are submitted via the API +
-  - iFrame embeddability is auto-detected +
-  - New lessons enter the Review Queue for human approval before going live+
  
-See [[developer:api|Collection API]] for technical details on the injection endpoint.+Claude scores how well each lesson matches each career role (0–100). 734 approved links are live, powering: 
 + 
 +  * "Lessons for this role" on IWTB role pages 
 +  * Personalised lesson recommendations 
 +  * Curriculum gap analysis per role 
 + 
 +===== AI-04: Editorial Governance Workspace ===== 
 + 
 +Navigate to ''/educator/ai-review-queue.php'' to: 
 + 
 +  * Review AI-generated lesson markup awaiting approval 
 +  * Accept, edit, or reject individual fields with field-level rollback 
 +  * Track OG / AI / Tags completion status per lesson 
 +  * Monitor which lessons are missing markup 
 + 
 +===== AI-06: EKOS Classification ===== 
 + 
 +The **European Key Occupations Standard** (EKOS) engine links lessons to: 
 + 
 +  * **UDC** (Universal Decimal Classification) subject categories 
 +  * **IFQ** (Irish Framework of Qualifications) levels 1–22 
 + 
 +Review and approve EKOS suggestions at ''/system/workflow_content.php''. 
 + 
 +===== Content Pipeline ===== 
 + 
 +The AI content pipeline generates complete lesson bundles from a topic list and injects them via the Lesson Injection API. The pipeline is fully **PHP-native** — no Python or external packages required on the server. 
 + 
 +**To queue a pipeline run:** 
 + 
 +  - Add topics to ''projects/topic_list.json'' 
 +  - The cron worker (''job_type = pipeline_ingest'') picks up the job 
 +  - Claude generates a complete lesson bundle per topic 
 +  - Lessons land in the review queue at ''/educator/lesson-review-queue.php'' 
 + 
 +**topic_list.json format:** 
 +<code json> 
 +[ 
 +  { 
 +    "topic": "How to send an email", 
 +    "context": "Digital skills for beginners", 
 +    "learning_objective": "Compose and send a basic email", 
 +    "target_level": "beginner", 
 +    "materials": ["https://youtube.com/watch?v=example"], 
 +    "notes": "Prefer practical framing." 
 +  } 
 +] 
 +</code> 
 + 
 +===== Lesson Review Queue ===== 
 + 
 +After pipeline ingestion, AI-generated lessons land in the **Lesson Review Queue** at ''/educator/lesson-review-queue.php'': 
 + 
 +  * Lesson title and summary 
 +  * OG / AI / Tags completion indicators 
 +  * Source (ai_pipeline, manual, etc.) 
 +  * Batch ID for traceability 
 + 
 +===== Technical Reference ===== 
 + 
 +  * Model: ''claude-haiku-4-5-20251001'' (configurable via ''ANTHROPIC_MODEL'' in env.php) 
 +  * Prompt caching enabled — ephemeral cache blocks reduce cost on repeated runs 
 +  * PHP client: ''cron/lib/claude_client.php'' — reusable for other cron jobs 
 +  * API key: ''ANTHROPIC_API_KEY'' in ''lib/const/env.php''