Skip to Content

Wiki source code of Workflows

Version 1.4 by Robert Schaub on 2026/01/20 20:25
Show last authors
1 = Workflows =
2
3 **Version:** 0.9.70 **Last Updated:** December 21, 2025 **Status:** CORRECTED - Automation Philosophy Consistent This page describes FactHarbor's core workflows with the automation-first philosophy. == 1. Core Workflow Principles == * **Automation First:** 90%+ content published automatically
4
5 * **No Approval Bottlenecks:** No centralized review queues
6 * **Quality Gates:** Automated validation before publication
7 * **Sampling Audits:** Pattern analysis for system improvement
8 * **Transparent Confidence:** All outputs labeled with confidence scores == 2. Claim Submission Workflow == === 2.1 Claim Extraction === When users submit content (text, articles, web pages), FactHarbor first extracts individual verifiable claims: **Input Types:**
9 * Single claim: "The Earth is flat"
10 * Text with multiple claims: "Climate change is accelerating. Sea levels rose 3mm in 2023. Arctic ice decreased 13% annually."
11 * URLs: Web pages analyzed for factual claims **Extraction Process:**
12 * LLM analyzes submitted content
13 * Identifies distinct, verifiable factual claims
14 * Separates claims from opinions, questions, or commentary
15 * Each claim becomes independent for processing **Output:**
16 * List of claims with context
17 * Each claim assigned unique ID
18 * Original context preserved for reference This extraction ensures:
19 * Each claim receives focused analysis
20 * Multiple claims in one submission are all processed
21 * Claims are properly isolated for independent verification
22 * Context is preserved for accurate interpretation **Flow:**
23 ```
24 User submits → Duplicate detection → Categorization → Processing queue → User receives ID
25 ``` **Timeline:** Seconds **No approval needed:** Instant processing == 3. Automated Analysis Workflow == **Complete Pipeline:** ```
26 Claim from queue
27 ↓
28 Evidence gathering (AKEL)
29 ↓
30 Source evaluation (track record check)
31 ↓
32 Scenario generation
33 ↓
34 Verdict synthesis
35 ↓
36 Risk assessment
37 ↓
38 Quality gates validation
39 ↓
40 Decision: PUBLISH or BLOCK
41 ``` **Timeline:** 10-30 seconds **Automation Rate:** 90%+ published automatically === 3.1 Quality Gates Decision == **Gate Validation:**
42
43 1. Gate 1: Source Quality ✓
44 2. Gate 2: Contradiction Search ✓
45 3. Gate 3: Uncertainty Quantification ✓
46 4. Gate 4: Structural Integrity ✓ **If ALL gates PASS:**
47 → **Publish immediately** (Mode 2: AI-Generated)
48 → Apply appropriate risk tier label
49 → Display confidence score
50 → Make available for sampling audit **If ANY gate FAILS:**
51 → **Block publication** (Mode 1: Draft-Only)
52 → Log failure reason
53 → Analyze failure pattern
54 → Queue system improvement task
55 → May re-process after improvements **CRITICAL:** No human approval step - gates are automated. == 4. Publication Workflow == **V0.9.70 CLARIFIED:** Risk tiers affect LABELS and AUDIT FREQUENCY, NOT approval requirements. === Standard Flow (90%+) === ```
56 Pass quality gates
57 ↓
58 Determine risk tier (A/B/C)
59 ↓
60 Apply appropriate labels
61 ↓
62 PUBLISH IMMEDIATELY
63 ↓
64 Add to audit sampling pool
65 ``` **No delays, no approval queues** === High-Risk Content (Tier A - <10%) === **V0.9.70 CORRECTION:** ```
66 Pass quality gates
67 ↓
68 Identified as Tier A (medical/legal/safety)
69 ↓
70 PUBLISH IMMEDIATELY with prominent warnings
71 ↓
72 Higher sampling audit frequency (50%)
73 ``` **What changed from V0.9.69:**
74 - ❌ REMOVED: "Risk > 80% → Moderator review"
75 - ✅ ADDED: "Risk > 80% → Publish with WARNING labels" **Philosophy:** Publish with strong warnings, monitor closely through sampling. **Warning Labels for Tier A:**
76 ```
77 ⚠️ HIGH-IMPACT TOPIC
78 AI-Generated Analysis This claim involves [medical/legal/financial/safety] topics.
79 - Confidence: [X]%
80 - Last Updated: [timestamp]
81 - This is NOT professional advice
82 - Consult qualified professionals for decisions [View Evidence] [See Methodology] [Report Issue]
83 ``` === Low Quality Content (<10%) === ```
84 FAIL quality gates
85 ↓
86 Confidence < threshold OR structural issues
87 ↓
88 BLOCK (Mode 1: Draft-Only)
89 ↓
90 Log failure patterns
91 ↓
92 Queue for system improvement
93 ``` **NOT:** Send for human review **IS:** Improve prompts/algorithms based on failure patterns == 5. User Contribution Workflow == **Philosophy:** Wikipedia-style immediate application + audit trail ```
94 Contributor edits published content
95 ↓
96 System validates (basic checks)
97 ↓
98 Applied IMMEDIATELY
99 ↓
100 Logged in version history
101 ↓
102 Reputation earned
103 ↓
104 May be selected for sampling audit
105 ``` **No approval required:** Changes apply instantly **Quality control:** Through sampling audits and reputation system **New contributors** (<50 reputation): Limited to minor edits == 6. Sampling Audit Workflow == **Purpose:** Improve system quality through pattern analysis === 6.1 Selection Process === ```
106 Published content
107 ↓
108 Stratified sampling (by risk tier, confidence, traffic)
109 ↓
110 Selected for audit (Tier A: 50%, B: 20%, C: 5%)
111 ↓
112 Added to audit queue
113 ``` === 6.2 Audit Execution === ```
114 Auditor receives sample
115 ↓
116 Reviews against quality standards
117 ↓
118 Identifies issues/patterns
119 ↓
120 Logs findings
121 ↓
122 System improvement tasks created
123 ``` **What auditors DO:**
124
125 * ✅ Analyze patterns across multiple outputs
126 * ✅ Identify systematic issues
127 * ✅ Recommend algorithm/prompt improvements
128 * ✅ Track accuracy trends **What auditors DON'T DO:**
129 * ❌ Approve individual outputs before publication
130 * ❌ Manually fix individual outputs
131 * ❌ Act as gatekeepers
132 * ❌ Override quality gates === 6.3 Improvement Loop === ```
133 Audit findings aggregated
134 ↓
135 Patterns identified
136 ↓
137 System improvements proposed
138 ↓
139 Implemented and tested
140 ↓
141 Deployed
142 ↓
143 Metrics monitored
144 ``` **Examples of Improvements:**
145 * Refine evidence search queries
146 * Adjust source reliability weights
147 * Enhance contradiction detection
148 * Improve claim extraction prompts
149 * Recalibrate risk tier thresholds == 7. Flagging Workflow == **Two types of flags:** === 7.1 Quality Issues === ```
150 User flags quality issue
151 ↓
152 Categorized automatically
153 ↓
154 Added to sampling audit pool (priority)
155 ↓
156 Pattern analysis
157 ↓
158 System improvement if pattern found
159 ``` **NOT:** Manual correction of individual claim **IS:** Improve system to prevent similar issues === 7.2 Abuse/Spam === ```
160 User flags abuse/spam
161 ↓
162 Automated pre-moderation check
163 ↓
164 Moderator review (if needed)
165 ↓
166 Action taken (hide/ban)
167 ``` **Moderator role:** Handle abuse/spam, NOT content quality == 8. Moderation Workflow == **V0.9.70 CLARIFIED:** Moderators handle ABUSE, not content quality === 8.1 Content Moderation (Abuse/Spam) === **Moderator Queue Contains:**
168 * Flagged abusive content
169 * Spam detection alerts
170 * Harassment reports
171 * Privacy violations
172 * Terms of service violations **Moderator Actions:**
173 * Hide abusive content
174 * Ban repeat offenders
175 * Handle appeals
176 * Escalate to governing team **Moderators DO NOT:**
177 * ❌ Approve content for publication
178 * ❌ Review content quality before publication
179 * ❌ Act as editorial gatekeepers
180 * ❌ Manually fix AI outputs === 8.2 Appeal Process === ```
181 User disagrees with moderation
182 ↓
183 Appeals to different moderator
184 ↓
185 If still disagrees, escalates to Governing Team
186 ↓
187 Governing Team decision (final)
188 ``` == 9. Time Evolution Workflow == **Automatic Re-evaluation:** ```
189 Published claim
190 ↓
191 Monitoring for triggers: - New evidence published - Source retractions - Significant events - Scheduled review ↓
192 Trigger detected
193 ↓
194 AKEL re-processes claim
195 ↓
196 Quality gates validate
197 ↓
198 If verdict changes: Correction workflow
199 ↓
200 If passes: Update published analysis
201 ``` **Correction Workflow (New in V0.9.70):** ```
202 Verdict changed significantly
203 ↓
204 Generate correction notice
205 ↓
206 Publish correction banner (30 days)
207 ↓
208 Update corrections log
209 ↓
210 Notify users (email, RSS, API)
211 ↓
212 Update ClaimReview schema
213 ``` == 10. Contributor Journey == 1. **Visitor** – Explores platform, reads documentation
214 2. **New Contributor** – Submits first improvements (typo fixes, clarifications)
215 3. **Contributor** – Contributes regularly, follows conventions
216 4. **Trusted Contributor** – Track record of quality work
217 5. **Reviewer** – Participates in sampling audits (pattern analysis)
218 6. **Moderator** – Handles abuse/spam (not content quality)
219 7. **Expert** (optional) – Provides domain expertise for contested claims **All contributions apply immediately** - no approval workflow == 11. Related Pages == * [[AKEL>>Archive.FactHarbor.Specification.AI Knowledge Extraction Layer (AKEL).WebHome]] - AI processing system
220 * [[Architecture>>FactHarbor.Specification.Architecture.WebHome]] - System architecture
221 * [[Requirements>>FactHarbor.Specification.Requirements.WebHome]] - Requirements and roles
222 * [[Decision Processes>>FactHarbor.Organisation.Decision-Processes.WebHome]] - Governance **V0.9.70 CHANGES:** **REMOVED:**
223 - ❌ "High Risk → Moderator review" (was approval workflow)
224 - ❌ "Review queue" language for publication
225 - ❌ Any implication that moderators approve content quality **ADDED/CLARIFIED:**
226 - ✅ Risk tiers affect warnings and audit frequency, NOT approval
227 - ✅ High-risk content publishes immediately with prominent warnings
228 - ✅ Quality gate failures → Block + improve system (not human review)
229 - ✅ Clear distinction: Sampling audits (improvement) vs. Content moderation (abuse)
230 - ✅ Moderator role clarified: Abuse only, NOT content quality
231 - ✅ User contributions apply immediately (Wikipedia model)
232 - ✅ Correction workflow for significant verdict changes
233 - ✅ Time evolution and re-evaluation workflow