Skip to Content
Last modified by Robert Schaub on 2025/12/24 20:33
From version 1.3
edited by Robert Schaub
on 2025/12/16 20:40
Change comment: Renamed back-links.
To version 1.2
edited by Robert Schaub
on 2025/12/16 20:40
Change comment: Update document after refactoring.

Summary

Details

Page properties
Content
... ... @@ -29,6 +29,7 @@
29 29  * Re-evaluation proposal generation
30 30  * Cross-node embedding exchange (Release 1.0+)
31 31  
32 +
32 32  == 2. Components ==
33 33  
34 34  * **AKEL Orchestrator** – central coordinator
... ... @@ -42,10 +42,10 @@
42 42  * **Embedding Handler** (Release 1.0+)
43 43  * **Federation Sync Adapter** (Release 1.0+)
44 44  
46 +
45 45  == 3. Inputs and Outputs ==
46 46  
47 47  === 3.1 Inputs ===
48 -
49 49  * User-submitted claims or evidence
50 50  * Uploaded documents
51 51  * URLs or citations
... ... @@ -53,7 +53,6 @@
53 53  * Embeddings (from local or federated peers)
54 54  
55 55  === 3.2 Outputs (publication mode varies by risk tier) ===
56 -
57 57  * ClaimVersion (draft or AI-generated)
58 58  * ScenarioVersion (draft or AI-generated)
59 59  * EvidenceVersion (summary + metadata, draft or AI-generated)
... ... @@ -63,6 +63,7 @@
63 63  * Re-evaluation proposals
64 64  * Updated embeddings
65 65  
66 +
66 66  == 4. Publication Modes ==
67 67  
68 68  AKEL content is published according to three modes:
... ... @@ -70,7 +70,6 @@
70 70  === 4.1 Mode 1: Draft-Only (Never Public) ===
71 71  
72 72  **Used for:**
73 -
74 74  * Failed quality gate checks
75 75  * Sensitive topics flagged for expert review
76 76  * Unclear scope or missing critical sources
... ... @@ -81,7 +81,6 @@
81 81  === 4.2 Mode 2: Published as AI-Generated (No Prior Human Review) ===
82 82  
83 83  **Requirements:**
84 -
85 85  * All automated quality gates passed (see below)
86 86  * Risk tier permits AI-draft publication (Tier B or C)
87 87  * Contradiction search completed successfully
... ... @@ -95,7 +95,6 @@
95 95  ```
96 96  
97 97  **User actions:**
98 -
99 99  * Browse and read content
100 100  * Request human review (escalates to review queue)
101 101  * Flag for expert attention
... ... @@ -103,7 +103,6 @@
103 103  === 4.3 Mode 3: Published as Human-Reviewed ===
104 104  
105 105  **Requirements:**
106 -
107 107  * Human reviewer or domain expert validated
108 108  * All quality gates passed
109 109  * Visible "Human-Reviewed" mark with reviewer role and timestamp
... ... @@ -125,13 +125,12 @@
125 125  **Domains:** Medical, legal, elections, safety/security, major reputational harm
126 126  
127 127  **Publication policy:**
128 -
129 129  * Human review REQUIRED before "Human-Reviewed" status
130 130  * AI-generated content MAY be published but:
131 -** Clearly flagged as AI-draft with prominent disclaimer
132 -** May have limited visibility
133 -** Auto-escalated to expert review queue
134 -** User warnings displayed
127 + ** Clearly flagged as AI-draft with prominent disclaimer
128 + ** May have limited visibility
129 + ** Auto-escalated to expert review queue
130 + ** User warnings displayed
135 135  
136 136  **Audit rate:** Recommendation: 30-50% of published AI-drafts sampled in first 6 months
137 137  
... ... @@ -140,7 +140,6 @@
140 140  **Domains:** Contested public policy, complex science, causality claims, significant financial impact
141 141  
142 142  **Publication policy:**
143 -
144 144  * AI-draft CAN publish immediately with clear labeling
145 145  * Sampling audits conducted (see Audit System below)
146 146  * High-engagement items auto-escalated to expert review
... ... @@ -153,7 +153,6 @@
153 153  **Domains:** Definitions, simple factual lookups with strong primary sources, historical facts, established scientific consensus
154 154  
155 155  **Publication policy:**
156 -
157 157  * AI-draft default publication mode
158 158  * Sampling audits sufficient
159 159  * Community flagging available
... ... @@ -167,7 +167,6 @@
167 167  All AI-generated content must pass these automated checks before Mode 2 publication:
168 168  
169 169  === 6.1 Gate 1: Source Quality ===
170 -
171 171  * Primary sources identified and accessible
172 172  * Source reliability scored against whitelist
173 173  * Citation completeness verified
... ... @@ -184,7 +184,6 @@
184 184  * **Bubble detection** – Conspiracy theories, echo chambers, ideologically isolated sources
185 185  
186 186  **Search coverage requirements:**
187 -
188 188  * Academic literature (BOTH supporting AND opposing views)
189 189  * Reputable media across diverse political/ideological perspectives
190 190  * Official contradictions (retractions, corrections, updates, amendments)
... ... @@ -192,7 +192,6 @@
192 192  * Cross-cultural and international perspectives
193 193  
194 194  **Search must actively avoid algorithmic bubbles:**
195 -
196 196  * Deliberately seek opposing viewpoints
197 197  * Check for echo chamber patterns in source clusters
198 198  * Identify tribal or ideological source clustering
... ... @@ -200,7 +200,6 @@
200 200  * Verify diversity of perspectives represented
201 201  
202 202  **Outcomes:**
203 -
204 204  * **Strong counter-evidence found** → Auto-escalate to Tier B or draft-only mode
205 205  * **Significant uncertainty detected** → Require uncertainty disclosure in verdict
206 206  * **Bubble indicators present** → Flag for expert review and human validation
... ... @@ -207,7 +207,6 @@
207 207  * **Limited perspective diversity** → Expand search or flag for human review
208 208  
209 209  === 6.3 Gate 3: Uncertainty Quantification ===
210 -
211 211  * Confidence scores calculated for all claims and verdicts
212 212  * Limitations explicitly stated
213 213  * Data gaps identified and disclosed
... ... @@ -215,7 +215,6 @@
215 215  * Alternative scenarios considered
216 216  
217 217  === 6.4 Gate 4: Structural Integrity ===
218 -
219 219  * No hallucinations detected (fact-checking against sources)
220 220  * Logic chain valid and traceable
221 221  * References accessible and verifiable
... ... @@ -223,12 +223,12 @@
223 223  * Premises clearly stated
224 224  
225 225  **If any gate fails:**
226 -
227 227  * Content remains in draft-only mode
228 228  * Failure reason logged
229 229  * Human review required before publication
230 230  * Failure patterns analyzed for system improvement
231 231  
219 +
232 232  == 7. Audit System (Sampling-Based Quality Assurance) ==
233 233  
234 234  Instead of reviewing ALL AI output, FactHarbor implements stratified sampling audits:
... ... @@ -236,7 +236,6 @@
236 236  === 7.1 Sampling Strategy ===
237 237  
238 238  Audits prioritize:
239 -
240 240  * **Risk tier** (higher tiers get more frequent audits)
241 241  * **AI confidence score** (low confidence → higher sampling rate)
242 242  * **Traffic and engagement** (high-visibility content audited more)
... ... @@ -248,12 +248,11 @@
248 248  1. System selects content for audit based on sampling strategy
249 249  2. Human auditor reviews AI-generated content against quality standards
250 250  3. Auditor validates or corrects:
251 -
252 -* Claim extraction accuracy
253 -* Scenario appropriateness
254 -* Evidence relevance and interpretation
255 -* Verdict reasoning
256 -* Contradiction search completeness
238 + * Claim extraction accuracy
239 + * Scenario appropriateness
240 + * Evidence relevance and interpretation
241 + * Verdict reasoning
242 + * Contradiction search completeness
257 257  4. Audit outcome recorded (pass/fail + detailed feedback)
258 258  5. Failed audits trigger immediate content review
259 259  6. Audit results feed back into system improvement
... ... @@ -261,7 +261,6 @@
261 261  === 7.3 Feedback Loop (Continuous Improvement) ===
262 262  
263 263  Audit outcomes systematically improve:
264 -
265 265  * **Query templates** – Refined based on missed evidence patterns
266 266  * **Retrieval source weights** – Adjusted for accuracy and reliability
267 267  * **Contradiction detection heuristics** – Enhanced to catch missed counter-evidence
... ... @@ -276,9 +276,10 @@
276 276  * System improvements documented
277 277  * Community can view aggregate audit performance
278 278  
264 +
279 279  == 8. Architecture Overview ==
280 280  
281 -{{include reference="Test.FactHarbor V0\.9\.23 Lost Data.Specification.Diagrams.AKEL Architecture.WebHome"}}
267 +{{include reference="Test.FactHarborV09.Specification.Diagrams.AKEL Architecture.WebHome"}}
282 282  
283 283  
284 284  == 9. AKEL and Federation ==
... ... @@ -333,4 +333,4 @@
333 333  * [[Requirements (Roles)>>Test.FactHarborV09.Specification.Requirements.WebHome]]
334 334  * [[Workflows>>Test.FactHarborV09.Specification.Workflows.WebHome]]
335 335  * [[Governance>>Test.FactHarborV09.Organisation.Governance]]
336 -{{/include}}
322 +