Skip to Content

Changes for page Workflows

Last modified by Robert Schaub on 2025/12/24 20:35
From version 1.3
edited by Robert Schaub
on 2025/12/16 20:40
Change comment: Updated parent field.
To version 1.1
edited by Robert Schaub
on 2025/12/16 20:34
Change comment: Imported from XAR

Summary

Details

Page properties
Parent
... ... @@ -1,1 +1,1 @@
1 -xwiki:Test.FactHarbor V0\.9\.23 Lost Data.Specification.WebHome
1 +Test.FactHarborV09.Specification.WebHome
Content
... ... @@ -20,7 +20,6 @@
20 20  **Actor**: Contributor or AKEL
21 21  
22 22  **Actions**:
23 -
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)
... ... @@ -30,7 +30,6 @@
30 30  === 2.2 Step 2: AKEL Processing ===
31 31  
32 32  **Automated Steps**:
33 -
34 34  1. Claim extraction and normalization
35 35  2. Classification (domain, type, evaluability)
36 36  3. Risk tier assignment (A/B/C suggested)
... ... @@ -44,7 +44,6 @@
44 44  === 2.3 Step 3: Quality Gate Checkpoint ===
45 45  
46 46  **Gates Evaluated**:
47 -
48 48  * Source quality
49 49  * Contradiction search completion
50 50  * Uncertainty quantification
... ... @@ -51,7 +51,6 @@
51 51  * Structural integrity
52 52  
53 53  **Outcomes**:
54 -
55 55  * **All gates pass** → Proceed to Mode 2 publication (if Tier B or C)
56 56  * **Any gate fails** → Mode 1 (Draft), flag for human review
57 57  * **Tier A** → Mode 2 with warnings + auto-escalate to expert queue
... ... @@ -59,19 +59,16 @@
59 59  === 2.4 Step 4: Publication (Risk-Tier Dependent) ===
60 60  
61 61  **Tier C (Low Risk)**:
62 -
63 63  * **Direct to Mode 2**: AI-generated, public, clearly labeled
64 64  * User can request human review
65 65  * Sampling audit applies
66 66  
67 67  **Tier B (Medium Risk)**:
68 -
69 69  * **Direct to Mode 2**: AI-generated, public, clearly labeled
70 70  * Higher audit sampling rate
71 71  * High-engagement content may auto-escalate
72 72  
73 73  **Tier A (High Risk)**:
74 -
75 75  * **Mode 2 with warnings**: AI-generated, public, prominent disclaimers
76 76  * **Auto-escalated** to expert review queue
77 77  * User warnings displayed
... ... @@ -80,7 +80,6 @@
80 80  === 2.5 Step 5: Human Review (Optional for B/C, Escalated for A) ===
81 81  
82 82  **Triggers**:
83 -
84 84  * User requests review
85 85  * Audit flags issues
86 86  * High engagement (Tier B)
... ... @@ -87,7 +87,6 @@
87 87  * Automatic (Tier A)
88 88  
89 89  **Process**:
90 -
91 91  1. Reviewer/Expert examines claim
92 92  2. Validates quality gates
93 93  3. Checks contradiction search results
... ... @@ -95,23 +95,21 @@
95 95  5. Decision: Approve, Request Changes, or Reject
96 96  
97 97  **Outcomes**:
98 -
99 99  * **Approved** → Mode 3 (Human-Reviewed)
100 100  * **Changes Requested** → Back to contributor or AKEL for revision
101 101  * **Rejected** → Rejected status with reasoning
102 102  
93 +
103 103  == 3. Scenario Creation Workflow ==
104 104  
105 105  === 3.1 Step 1: Scenario Generation ===
106 106  
107 107  **Automated (AKEL)**:
108 -
109 109  * Generate scenarios for claim
110 110  * Define boundaries, assumptions, context
111 111  * Identify evaluation methods
112 112  
113 113  **Manual (Expert/Reviewer)**:
114 -
115 115  * Create custom scenarios
116 116  * Refine AKEL-generated scenarios
117 117  * Add domain-specific nuances
... ... @@ -119,7 +119,6 @@
119 119  === 3.2 Step 2: Scenario Validation ===
120 120  
121 121  **Quality Checks**:
122 -
123 123  * Completeness (definitions, boundaries, assumptions clear)
124 124  * Relevance to claim
125 125  * Evaluability
... ... @@ -126,7 +126,6 @@
126 126  * No circular logic
127 127  
128 128  **Risk Tier Assignment**:
129 -
130 130  * Inherits from parent claim
131 131  * Can be overridden by expert if scenario increases/decreases risk
132 132  
... ... @@ -133,21 +133,19 @@
133 133  === 3.3 Step 3: Scenario Publication ===
134 134  
135 135  **Mode 2 (AI-Generated)**:
136 -
137 137  * Tier B/C scenarios can publish immediately
138 138  * Subject to sampling audits
139 139  
140 140  **Mode 1 (Draft)**:
141 -
142 142  * Tier A scenarios default to draft
143 143  * Require expert validation for Mode 2 or Mode 3
144 144  
130 +
145 145  == 4. Evidence Evaluation Workflow ==
146 146  
147 147  === 4.1 Step 1: Evidence Search & Retrieval ===
148 148  
149 149  **AKEL Actions**:
150 -
151 151  * Search academic databases, reputable media
152 152  * **Mandatory contradiction search** (counter-evidence, reservations)
153 153  * Extract metadata (author, date, publication, methodology)
... ... @@ -154,7 +154,6 @@
154 154  * Assess source reliability
155 155  
156 156  **Quality Requirements**:
157 -
158 158  * Primary sources preferred
159 159  * Diverse perspectives included
160 160  * Echo chambers flagged
... ... @@ -163,7 +163,6 @@
163 163  === 4.2 Step 2: Evidence Summarization ===
164 164  
165 165  **AKEL Generates**:
166 -
167 167  * Summary of evidence
168 168  * Relevance assessment
169 169  * Reliability score
... ... @@ -175,7 +175,6 @@
175 175  === 4.3 Step 3: Evidence Review ===
176 176  
177 177  **Reviewer/Expert Validates**:
178 -
179 179  * Accuracy of summaries
180 180  * Appropriateness of sources
181 181  * Completeness of contradiction search
... ... @@ -182,17 +182,16 @@
182 182  * Reliability assessments
183 183  
184 184  **Outcomes**:
185 -
186 186  * **Mode 2**: Evidence summaries published as AI-generated
187 187  * **Mode 3**: After human validation
188 188  * **Mode 1**: Failed quality checks or pending expert review
189 189  
171 +
190 190  == 5. Verdict Generation Workflow ==
191 191  
192 192  === 5.1 Step 1: Verdict Computation ===
193 193  
194 194  **AKEL Computes**:
195 -
196 196  * Verdict across scenarios
197 197  * Confidence scores
198 198  * Uncertainty quantification
... ... @@ -200,7 +200,6 @@
200 200  * Limitations
201 201  
202 202  **Inputs**:
203 -
204 204  * Claim text
205 205  * Scenario definitions
206 206  * Evidence assessments
... ... @@ -209,13 +209,11 @@
209 209  === 5.2 Step 2: Verdict Validation ===
210 210  
211 211  **Quality Gates**:
212 -
213 213  * All four gates apply (source, contradiction, uncertainty, structure)
214 214  * Reasoning chain must be traceable
215 215  * Assumptions must be explicit
216 216  
217 217  **Risk Tier Check**:
218 -
219 219  * Tier A: Always requires expert validation for Mode 3
220 220  * Tier B: Mode 2 allowed, audit sampling
221 221  * Tier C: Mode 2 default
... ... @@ -223,7 +223,6 @@
223 223  === 5.3 Step 3: Verdict Publication ===
224 224  
225 225  **Mode 2 (AI-Generated Verdict)**:
226 -
227 227  * Clear labeling with confidence scores
228 228  * Uncertainty disclosure
229 229  * Links to reasoning trail
... ... @@ -230,17 +230,16 @@
230 230  * User can request expert review
231 231  
232 232  **Mode 3 (Expert-Validated Verdict)**:
233 -
234 234  * Human reviewer/expert stamp
235 235  * Additional commentary (optional)
236 236  * Highest trust level
237 237  
214 +
238 238  == 6. Audit Workflow ==
239 239  
240 240  === 6.1 Step 1: Audit Sampling Selection ===
241 241  
242 242  **Stratified Sampling**:
243 -
244 244  * Risk tier priority (A > B > C)
245 245  * Low confidence scores
246 246  * High traffic content
... ... @@ -248,7 +248,6 @@
248 248  * User flags
249 249  
250 250  **Sampling Rates** (Recommendations):
251 -
252 252  * Tier A: 30-50%
253 253  * Tier B: 10-20%
254 254  * Tier C: 5-10%
... ... @@ -256,7 +256,6 @@
256 256  === 6.2 Step 2: Audit Execution ===
257 257  
258 258  **Auditor Actions**:
259 -
260 260  1. Review sampled AI-generated content
261 261  2. Validate quality gates were properly applied
262 262  3. Check contradiction search completeness
... ... @@ -264,7 +264,6 @@
264 264  5. Identify errors or hallucinations
265 265  
266 266  **Audit Outcome**:
267 -
268 268  * **Pass**: Content remains in Mode 2, logged as validated
269 269  * **Fail**: Content flagged for review, system improvement triggered
270 270  
... ... @@ -271,7 +271,6 @@
271 271  === 6.3 Step 3: Feedback Loop ===
272 272  
273 273  **System Improvements**:
274 -
275 275  * Failed audits analyzed for patterns
276 276  * AKEL parameters adjusted
277 277  * Quality gates refined
... ... @@ -278,17 +278,16 @@
278 278  * Risk tier assignments recalibrated
279 279  
280 280  **Transparency**:
281 -
282 282  * Audit statistics published periodically
283 283  * Patterns shared with community
284 284  * System improvements documented
285 285  
257 +
286 286  == 7. Mode Transition Workflow ==
287 287  
288 288  === 7.1 Mode 1 → Mode 2 ===
289 289  
290 290  **Requirements**:
291 -
292 292  * All quality gates pass
293 293  * Risk tier B or C (or A with warnings)
294 294  * Contradiction search completed
... ... @@ -298,7 +298,6 @@
298 298  === 7.2 Mode 2 → Mode 3 ===
299 299  
300 300  **Requirements**:
301 -
302 302  * Human reviewer/expert validation
303 303  * Quality standards confirmed
304 304  * For Tier A: Expert approval required
... ... @@ -309,29 +309,26 @@
309 309  === 7.3 Mode 3 → Mode 1 (Demotion) ===
310 310  
311 311  **Rare - Only if**:
312 -
313 313  * New evidence contradicts verdict
314 314  * Error discovered in reasoning
315 315  * Source retraction
316 316  
317 317  **Process**:
318 -
319 319  1. Content flagged for re-evaluation
320 320  2. Moved to draft (Mode 1)
321 321  3. Re-processed through workflow
322 322  4. Reason for demotion documented
323 323  
292 +
324 324  == 8. User Actions Across Modes ==
325 325  
326 326  === 8.1 On Mode 1 (Draft) Content ===
327 327  
328 328  **Contributors**:
329 -
330 330  * Edit their own drafts
331 331  * Submit for review
332 332  
333 333  **Reviewers/Experts**:
334 -
335 335  * View and comment
336 336  * Request changes
337 337  * Approve for Mode 2 or Mode 3
... ... @@ -339,7 +339,6 @@
339 339  === 8.2 On Mode 2 (AI-Generated) Content ===
340 340  
341 341  **All Users**:
342 -
343 343  * Read and use content
344 344  * Request human review
345 345  * Flag for expert attention
... ... @@ -346,7 +346,6 @@
346 346  * Provide feedback
347 347  
348 348  **Reviewers/Experts**:
349 -
350 350  * Validate for Mode 3 transition
351 351  * Edit and refine
352 352  * Adjust risk tier if needed
... ... @@ -354,20 +354,19 @@
354 354  === 8.3 On Mode 3 (Human-Reviewed) Content ===
355 355  
356 356  **All Users**:
357 -
358 358  * Read with highest confidence
359 359  * Still can flag if new evidence emerges
360 360  
361 361  **Reviewers/Experts**:
362 -
363 363  * Update if needed
364 364  * Trigger re-evaluation if new evidence
365 365  
329 +
366 366  == 9. Diagram References ==
367 367  
368 368  === 9.1 Claim and Scenario Lifecycle (Overview) ===
369 369  
370 -{{include reference="Test.FactHarbor V0\.9\.23 Lost Data.Organisation.Diagrams.Claim and Scenario Lifecycle (Overview).WebHome"}}
334 +{{include reference="Test.FactHarborV09.Organisation.Diagrams.Claim and Scenario Lifecycle (Overview).WebHome"}}
371 371  
372 372  === 9.2 Claim and Scenario Workflow ===
373 373  
... ... @@ -392,4 +392,4 @@
392 392  * [[Automation>>Test.FactHarborV09.Specification.Automation.WebHome]]
393 393  * [[Requirements (Roles)>>Test.FactHarborV09.Specification.Requirements.WebHome]]
394 394  * [[Governance>>Test.FactHarborV09.Organisation.Governance]]
395 -{{/include}}
359 +