Last modified by Robert Schaub on 2025/12/22 13:50

From version 1.5
edited by Robert Schaub
on 2025/12/22 13:49
Change comment: Renamed back-links.
To version 1.2
edited by Robert Schaub
on 2025/12/22 13:49
Change comment: Update document after refactoring.

Summary

Details

Page properties
Content
... ... @@ -12,12 +12,12 @@
12 12  **Key Innovation:** Quality validation BEFORE publication, not after
13 13  
14 14  **What We're Proving:**
15 -
16 16  * AKEL can reliably extract factual claims from articles
17 17  * AKEL can generate credible verdicts with proper evidence
18 18  * Quality gates prevent hallucinations and low-confidence outputs
19 19  * Fully automated approach is viable
20 20  
20 +
21 21  == 2. Scope ==
22 22  
23 23  === In Scope ===
... ... @@ -40,6 +40,7 @@
40 40  * A/B testing
41 41  * Gates 2 & 3 (Evidence relevance, Scenario coherence)
42 42  
43 +
43 43  == 3. Requirements ==
44 44  
45 45  === 3.1 NFR11: Quality Assurance Framework (POC1 Lite Version) ===
... ... @@ -56,7 +56,6 @@
56 56  **Purpose:** Ensure extracted claims are factual assertions, not opinions or predictions
57 57  
58 58  **Validation Checks:**
59 -
60 60  1. **Factual Statement Test:** Can this be verified with evidence?
61 61  2. **Opinion Detection:** Contains hedging language? ("I think", "probably", "best", "worst")
62 62  3. **Specificity Score:** Contains concrete details? (names, numbers, dates, locations)
... ... @@ -63,13 +63,14 @@
63 63  4. **Future Prediction Test:** Makes claims about future events?
64 64  
65 65  **Pass Criteria:**
66 -{{code}}- isFactual: true
66 +{{code}}
67 +- isFactual: true
67 67  - opinionScore: ≤ 0.3
68 68  - specificityScore: ≥ 0.3
69 -- claimType: FACTUAL{{/code}}
70 +- claimType: FACTUAL
71 +{{/code}}
70 70  
71 71  **Action if Failed:**
72 -
73 73  * Flag as "Non-verifiable: Opinion/Prediction/Ambiguous"
74 74  * Do NOT generate scenarios or verdicts
75 75  * Display explanation to user
... ... @@ -82,7 +82,6 @@
82 82  **Purpose:** Only publish verdicts with sufficient evidence and confidence
83 83  
84 84  **Validation Checks:**
85 -
86 86  1. **Evidence Count:** Minimum 2 independent sources
87 87  2. **Source Quality:** Average reliability ≥ 0.6 (on 0-1 scale)
88 88  3. **Evidence Agreement:** % supporting vs. contradicting ≥ 0.6
... ... @@ -89,7 +89,8 @@
89 89  4. **Uncertainty Factors:** Count of hedging statements in reasoning
90 90  
91 91  **Confidence Tiers:**
92 -{{code}}HIGH (80-100%):
92 +{{code}}
93 +HIGH (80-100%):
93 93   - ≥3 sources
94 94   - ≥0.7 average quality
95 95   - ≥80% agreement
... ... @@ -103,10 +103,10 @@
103 103   - ≥2 sources BUT low quality/agreement
104 104  
105 105  INSUFFICIENT:
106 - - <2 sources → DO NOT PUBLISH{{/code}}
107 + - <2 sources → DO NOT PUBLISH
108 +{{/code}}
107 107  
108 108  **POC1 Publication Rule:**
109 -
110 110  * Minimum **MEDIUM** confidence required
111 111  * Blocked verdicts show "Insufficient Evidence" message
112 112  
... ... @@ -139,7 +139,6 @@
139 139  {{/code}}
140 140  
141 141  **Updated Verdict States:**
142 -
143 143  * PUBLISHED - Passed all gates
144 144  * INSUFFICIENT_EVIDENCE - Failed Gate 4
145 145  * NON_FACTUAL_CLAIM - Failed Gate 1
... ... @@ -146,6 +146,7 @@
146 146  * PROCESSING - In progress
147 147  * ERROR - System failure
148 148  
149 +
149 149  === 3.3 Modified FR4: Analysis Summary (Enhanced) ===
150 150  
151 151  **Enhancement for POC1:**
... ... @@ -176,7 +176,6 @@
176 176  POC1 is considered **SUCCESSFUL** if:
177 177  
178 178  **✅ Functional:**
179 -
180 180  * Processes diverse test articles without crashes
181 181  * Generates verdicts for all factual claims
182 182  * Blocks all non-factual claims (0% pass through)
... ... @@ -183,7 +183,6 @@
183 183  * Blocks all insufficient-evidence verdicts (0% with <2 sources)
184 184  
185 185  **✅ Quality:**
186 -
187 187  * Hallucination rate <10% (manual verification)
188 188  * 0 verdicts with <2 sources published
189 189  * 0 opinion statements published as facts
... ... @@ -190,18 +190,17 @@
190 190  * Average quality score ≥7.0/10
191 191  
192 192  **✅ Performance:**
193 -
194 194  * Processing time reasonable for POC demonstration
195 195  * Quality gates execute efficiently
196 196  * UI displays results clearly
197 197  
198 198  **✅ Learnings:**
199 -
200 200  * Identified prompt engineering improvements
201 201  * Documented AKEL strengths/weaknesses
202 202  * Validated threshold values
203 203  * Clear path to POC2 defined
204 204  
202 +
205 205  == 5. Decision Gates ==
206 206  
207 207  **POC1 → POC2 Decision:**
... ... @@ -232,7 +232,6 @@
232 232  {{/code}}
233 233  
234 234  **POC1 Acceptable Simplifications:**
235 -
236 236  * Single AKEL call (not multi-component pipeline)
237 237  * No scenarios (implicit in verdicts)
238 238  * Basic evidence linking
... ... @@ -239,16 +239,18 @@
239 239  * 2 gates instead of 4
240 240  * No review queue
241 241  
242 -**See:** [[Architecture>>Test.FactHarbor pre10 V0\.9\.70.Specification.Architecture.WebHome]] for details
239 +**See:** [[Architecture>>Test.FactHarbor.Specification.Architecture.WebHome]] for details
243 243  
244 244  
245 245  == Related Pages ==
246 246  
247 -* [[Roadmap Overview>>Test.FactHarbor pre10 V0\.9\.70.Roadmap.WebHome]] - All phases
248 -* [[POC2 Requirements>>Test.FactHarbor pre10 V0\.9\.70.Roadmap.POC2.WebHome]] - Next phase
244 +* [[Roadmap Overview>>Test.FactHarbor.Roadmap.WebHome]] - All phases
245 +* [[POC2 Requirements>>Test.FactHarbor.Roadmap.POC2.WebHome]] - Next phase
249 249  * [[Requirements>>Test.FactHarbor.Specification.Requirements.WebHome]] - Full system requirements
250 -* [[Architecture>>Test.FactHarbor pre10 V0\.9\.70.Specification.Architecture.WebHome]] - System architecture
247 +* [[Architecture>>Test.FactHarbor.Specification.Architecture.WebHome]] - System architecture
251 251  * [[NFR11 Full Specification>>Test.FactHarbor.Specification.Requirements.WebHome#NFR11]] - Complete quality framework
252 252  
250 +
253 253  **Document Status:** ✅ POC1 Specification Complete - Ready for Implementation
254 254  **Version:** V0.9.70
253 +