Skip to Content

Changes for page Workflows

Last modified by Robert Schaub on 2025/12/24 20:34
From version 6.1
edited by Robert Schaub
on 2025/12/14 18:59
Change comment: Imported from XAR
To version 7.2
edited by Robert Schaub
on 2025/12/16 20:28
Change comment: Renamed back-links.

Summary

Details

Page properties
Content
... ... @@ -21,6 +21,7 @@
21 21  **Actor**: Contributor or AKEL
22 22  
23 23  **Actions**:
24 +
24 24  * Submit claim text
25 25  * Provide initial sources (optional for human contributors, mandatory for AKEL)
26 26  * System assigns initial AuthorType (Human or AI)
... ... @@ -30,6 +30,7 @@
30 30  === Step 2: AKEL Processing ===
31 31  
32 32  **Automated Steps**:
34 +
33 33  1. Claim extraction and normalization
34 34  2. Classification (domain, type, evaluability)
35 35  3. Risk tier assignment (A/B/C suggested)
... ... @@ -43,6 +43,7 @@
43 43  === Step 3: Quality Gate Checkpoint ===
44 44  
45 45  **Gates Evaluated**:
48 +
46 46  * Source quality
47 47  * Contradiction search completion
48 48  * Uncertainty quantification
... ... @@ -49,6 +49,7 @@
49 49  * Structural integrity
50 50  
51 51  **Outcomes**:
55 +
52 52  * **All gates pass** → Proceed to Mode 2 publication (if Tier B or C)
53 53  * **Any gate fails** → Mode 1 (Draft), flag for human review
54 54  * **Tier A** → Mode 2 with warnings + auto-escalate to expert queue
... ... @@ -56,16 +56,19 @@
56 56  === Step 4: Publication (Risk-Tier Dependent) ===
57 57  
58 58  **Tier C (Low Risk)**:
63 +
59 59  * **Direct to Mode 2**: AI-generated, public, clearly labeled
60 60  * User can request human review
61 61  * Sampling audit applies
62 62  
63 63  **Tier B (Medium Risk)**:
69 +
64 64  * **Direct to Mode 2**: AI-generated, public, clearly labeled
65 65  * Higher audit sampling rate
66 66  * High-engagement content may auto-escalate
67 67  
68 68  **Tier A (High Risk)**:
75 +
69 69  * **Mode 2 with warnings**: AI-generated, public, prominent disclaimers
70 70  * **Auto-escalated** to expert review queue
71 71  * User warnings displayed
... ... @@ -74,6 +74,7 @@
74 74  === Step 5: Human Review (Optional for B/C, Escalated for A) ===
75 75  
76 76  **Triggers**:
84 +
77 77  * User requests review
78 78  * Audit flags issues
79 79  * High engagement (Tier B)
... ... @@ -80,6 +80,7 @@
80 80  * Automatic (Tier A)
81 81  
82 82  **Process**:
91 +
83 83  1. Reviewer/Expert examines claim
84 84  2. Validates quality gates
85 85  3. Checks contradiction search results
... ... @@ -87,6 +87,7 @@
87 87  5. Decision: Approve, Request Changes, or Reject
88 88  
89 89  **Outcomes**:
99 +
90 90  * **Approved** → Mode 3 (Human-Reviewed)
91 91  * **Changes Requested** → Back to contributor or AKEL for revision
92 92  * **Rejected** → Rejected status with reasoning
... ... @@ -98,11 +98,13 @@
98 98  === Step 1: Scenario Generation ===
99 99  
100 100  **Automated (AKEL)**:
111 +
101 101  * Generate scenarios for claim
102 102  * Define boundaries, assumptions, context
103 103  * Identify evaluation methods
104 104  
105 105  **Manual (Expert/Reviewer)**:
117 +
106 106  * Create custom scenarios
107 107  * Refine AKEL-generated scenarios
108 108  * Add domain-specific nuances
... ... @@ -110,6 +110,7 @@
110 110  === Step 2: Scenario Validation ===
111 111  
112 112  **Quality Checks**:
125 +
113 113  * Completeness (definitions, boundaries, assumptions clear)
114 114  * Relevance to claim
115 115  * Evaluability
... ... @@ -116,6 +116,7 @@
116 116  * No circular logic
117 117  
118 118  **Risk Tier Assignment**:
132 +
119 119  * Inherits from parent claim
120 120  * Can be overridden by expert if scenario increases/decreases risk
121 121  
... ... @@ -122,10 +122,12 @@
122 122  === Step 3: Scenario Publication ===
123 123  
124 124  **Mode 2 (AI-Generated)**:
139 +
125 125  * Tier B/C scenarios can publish immediately
126 126  * Subject to sampling audits
127 127  
128 128  **Mode 1 (Draft)**:
144 +
129 129  * Tier A scenarios default to draft
130 130  * Require expert validation for Mode 2 or Mode 3
131 131  
... ... @@ -136,6 +136,7 @@
136 136  === Step 1: Evidence Search & Retrieval ===
137 137  
138 138  **AKEL Actions**:
155 +
139 139  * Search academic databases, reputable media
140 140  * **Mandatory contradiction search** (counter-evidence, reservations)
141 141  * Extract metadata (author, date, publication, methodology)
... ... @@ -142,6 +142,7 @@
142 142  * Assess source reliability
143 143  
144 144  **Quality Requirements**:
162 +
145 145  * Primary sources preferred
146 146  * Diverse perspectives included
147 147  * Echo chambers flagged
... ... @@ -150,6 +150,7 @@
150 150  === Step 2: Evidence Summarization ===
151 151  
152 152  **AKEL Generates**:
171 +
153 153  * Summary of evidence
154 154  * Relevance assessment
155 155  * Reliability score
... ... @@ -161,6 +161,7 @@
161 161  === Step 3: Evidence Review ===
162 162  
163 163  **Reviewer/Expert Validates**:
183 +
164 164  * Accuracy of summaries
165 165  * Appropriateness of sources
166 166  * Completeness of contradiction search
... ... @@ -167,6 +167,7 @@
167 167  * Reliability assessments
168 168  
169 169  **Outcomes**:
190 +
170 170  * **Mode 2**: Evidence summaries published as AI-generated
171 171  * **Mode 3**: After human validation
172 172  * **Mode 1**: Failed quality checks or pending expert review
... ... @@ -178,6 +178,7 @@
178 178  === Step 1: Verdict Computation ===
179 179  
180 180  **AKEL Computes**:
202 +
181 181  * Verdict across scenarios
182 182  * Confidence scores
183 183  * Uncertainty quantification
... ... @@ -185,6 +185,7 @@
185 185  * Limitations
186 186  
187 187  **Inputs**:
210 +
188 188  * Claim text
189 189  * Scenario definitions
190 190  * Evidence assessments
... ... @@ -193,11 +193,13 @@
193 193  === Step 2: Verdict Validation ===
194 194  
195 195  **Quality Gates**:
219 +
196 196  * All four gates apply (source, contradiction, uncertainty, structure)
197 197  * Reasoning chain must be traceable
198 198  * Assumptions must be explicit
199 199  
200 200  **Risk Tier Check**:
225 +
201 201  * Tier A: Always requires expert validation for Mode 3
202 202  * Tier B: Mode 2 allowed, audit sampling
203 203  * Tier C: Mode 2 default
... ... @@ -205,6 +205,7 @@
205 205  === Step 3: Verdict Publication ===
206 206  
207 207  **Mode 2 (AI-Generated Verdict)**:
233 +
208 208  * Clear labeling with confidence scores
209 209  * Uncertainty disclosure
210 210  * Links to reasoning trail
... ... @@ -211,6 +211,7 @@
211 211  * User can request expert review
212 212  
213 213  **Mode 3 (Expert-Validated Verdict)**:
240 +
214 214  * Human reviewer/expert stamp
215 215  * Additional commentary (optional)
216 216  * Highest trust level
... ... @@ -222,6 +222,7 @@
222 222  === Step 1: Audit Sampling Selection ===
223 223  
224 224  **Stratified Sampling**:
252 +
225 225  * Risk tier priority (A > B > C)
226 226  * Low confidence scores
227 227  * High traffic content
... ... @@ -229,6 +229,7 @@
229 229  * User flags
230 230  
231 231  **Sampling Rates** (Recommendations):
260 +
232 232  * Tier A: 30-50%
233 233  * Tier B: 10-20%
234 234  * Tier C: 5-10%
... ... @@ -236,6 +236,7 @@
236 236  === Step 2: Audit Execution ===
237 237  
238 238  **Auditor Actions**:
268 +
239 239  1. Review sampled AI-generated content
240 240  2. Validate quality gates were properly applied
241 241  3. Check contradiction search completeness
... ... @@ -243,6 +243,7 @@
243 243  5. Identify errors or hallucinations
244 244  
245 245  **Audit Outcome**:
276 +
246 246  * **Pass**: Content remains in Mode 2, logged as validated
247 247  * **Fail**: Content flagged for review, system improvement triggered
248 248  
... ... @@ -249,6 +249,7 @@
249 249  === Step 3: Feedback Loop ===
250 250  
251 251  **System Improvements**:
283 +
252 252  * Failed audits analyzed for patterns
253 253  * AKEL parameters adjusted
254 254  * Quality gates refined
... ... @@ -255,6 +255,7 @@
255 255  * Risk tier assignments recalibrated
256 256  
257 257  **Transparency**:
290 +
258 258  * Audit statistics published periodically
259 259  * Patterns shared with community
260 260  * System improvements documented
... ... @@ -266,6 +266,7 @@
266 266  === Mode 1 → Mode 2 ===
267 267  
268 268  **Requirements**:
302 +
269 269  * All quality gates pass
270 270  * Risk tier B or C (or A with warnings)
271 271  * Contradiction search completed
... ... @@ -275,6 +275,7 @@
275 275  === Mode 2 → Mode 3 ===
276 276  
277 277  **Requirements**:
312 +
278 278  * Human reviewer/expert validation
279 279  * Quality standards confirmed
280 280  * For Tier A: Expert approval required
... ... @@ -285,11 +285,13 @@
285 285  === Mode 3 → Mode 1 (Demotion) ===
286 286  
287 287  **Rare - Only if**:
323 +
288 288  * New evidence contradicts verdict
289 289  * Error discovered in reasoning
290 290  * Source retraction
291 291  
292 292  **Process**:
329 +
293 293  1. Content flagged for re-evaluation
294 294  2. Moved to draft (Mode 1)
295 295  3. Re-processed through workflow
... ... @@ -302,10 +302,12 @@
302 302  === On Mode 1 (Draft) Content ===
303 303  
304 304  **Contributors**:
342 +
305 305  * Edit their own drafts
306 306  * Submit for review
307 307  
308 308  **Reviewers/Experts**:
347 +
309 309  * View and comment
310 310  * Request changes
311 311  * Approve for Mode 2 or Mode 3
... ... @@ -313,6 +313,7 @@
313 313  === On Mode 2 (AI-Generated) Content ===
314 314  
315 315  **All Users**:
355 +
316 316  * Read and use content
317 317  * Request human review
318 318  * Flag for expert attention
... ... @@ -319,6 +319,7 @@
319 319  * Provide feedback
320 320  
321 321  **Reviewers/Experts**:
362 +
322 322  * Validate for Mode 3 transition
323 323  * Edit and refine
324 324  * Adjust risk tier if needed
... ... @@ -326,10 +326,12 @@
326 326  === On Mode 3 (Human-Reviewed) Content ===
327 327  
328 328  **All Users**:
370 +
329 329  * Read with highest confidence
330 330  * Still can flag if new evidence emerges
331 331  
332 332  **Reviewers/Experts**:
375 +
333 333  * Update if needed
334 334  * Trigger re-evaluation if new evidence
335 335  
... ... @@ -337,10 +337,26 @@
337 337  
338 338  == Diagram References ==
339 339  
340 -{{include reference="FactHarbor.Organisation.Diagrams.Claim-Scenario-Lifecycle"/}}
383 +=== Claim and Scenario Lifecycle (Overview) ===
341 341  
342 -{{include reference="FactHarbor.Specification.Diagrams.Manual vs Automated matrix.WebHome"/}}
385 +{{include reference="FactHarbor.Archive.FactHarbor V0\.9\.23 Lost Data.Organisation.Diagrams.Claim and Scenario Lifecycle (Overview).WebHome"/}}
343 343  
387 +=== Claim and Scenario Workflow ===
388 +
389 +{{include reference="Test.FactHarborV09.Specification.Diagrams.Claim and Scenario Workflow.WebHome"/}}
390 +
391 +=== Evidence and Verdict Workflow ===
392 +
393 +{{include reference="Test.FactHarborV09.Specification.Diagrams.Evidence and Verdict Workflow.WebHome"/}}
394 +
395 +=== Quality and Audit Workflow ===
396 +
397 +{{include reference="Test.FactHarborV09.Specification.Diagrams.Quality and Audit Workflow.WebHome"/}}
398 +
399 +
400 +
401 +{{include reference="Test.FactHarborV09.Specification.Diagrams.Manual vs Automated matrix.WebHome"/}}
402 +
344 344  ----
345 345  
346 346  == Related Pages ==
... ... @@ -349,4 +349,3 @@
349 349  * [[Automation>>FactHarbor.Specification.Automation.WebHome]]
350 350  * [[Requirements (Roles)>>FactHarbor.Specification.Requirements.WebHome]]
351 351  * [[Governance>>FactHarbor.Organisation.Governance]]
352 -