Skip to Content

Wiki source code of Workflows

Last modified by Robert Schaub on 2026/02/08 08:26
Show last authors
1 = Workflows =
2
3 FactHarbor workflows are **simple, automated, focused on continuous improvement**.
4
5 == 1. Core Principles ==
6
7 * **Automated by default**: AI processes everything
8 * **Publish immediately**: No centralized approval (removed in V0.9.50)
9 * **Quality through monitoring**: Not gatekeeping
10 * **Fix systems, not data**: Errors trigger improvements
11 * **Human-in-loop**: Only for edge cases and abuse
12
13 == 2. Claim Submission Workflow ==
14
15 === 2.1 Claim Extraction ===
16
17 When users submit content (text, articles, web pages), FactHarbor first extracts individual verifiable claims:
18
19 **Input Types:**
20
21 * Single claim: "The Earth is flat"
22 * Text with multiple claims: "Climate change is accelerating. Sea levels rose 3mm in 2023. Arctic ice decreased 13% annually."
23 * URLs: Web pages analyzed for factual claims
24
25 **Extraction Process:**
26
27 * LLM analyzes submitted content
28 * Identifies distinct, verifiable factual claims
29 * Separates claims from opinions, questions, or commentary
30 * Each claim becomes independent for processing
31
32 **Output:**
33
34 * List of claims with context
35 * Each claim assigned unique ID
36 * Original context preserved for reference
37
38 This extraction ensures:
39
40 * Each claim receives focused analysis
41 * Multiple claims in one submission are all processed
42 * Claims are properly isolated for independent verification
43 * Context is preserved for accurate interpretation
44
45 ```
46 User submits → Duplicate detection → Categorization → Processing queue → User receives ID
47 ```
48 **Timeline**: Seconds
49 **No approval needed**
50
51 == 2.5 Claim and Scenario Workflow ==
52
53 {{include reference="Archive.FactHarbor 2026\.01\.20.Specification.Diagrams.Claim and Scenario Workflow.WebHome"/}}
54
55 == 3. Automated Analysis Workflow ==
56
57 ```
58 Claim from queue
59 ↓
60 Evidence gathering (AKEL)
61 ↓
62 Source evaluation (track record check)
63 ↓
64 Scenario generation
65 ↓
66 Verdict synthesis
67 ↓
68 Risk assessment
69 ↓
70 Quality gates (confidence > 40%? risk < 80%?)
71 ↓
72 Publish OR Flag for improvement
73 ```
74 **Timeline**: 10-30 seconds
75 **90%+ published automatically**
76
77 == 3.5 Evidence and Verdict Workflow ==
78
79 {{include reference="Archive.FactHarbor 2026\.01\.20.Specification.Diagrams.Evidence and Verdict Workflow.WebHome"/}}
80
81 == 4. Publication Workflow ==
82
83 **Standard (90%+)**: Pass quality gates → Publish immediately with confidence scores
84 **High Risk (<10%)**: Risk > 80% → Moderator review
85 **Low Quality**: Confidence < 40% → Improvement queue → Re-process
86
87 == 5. User Contribution Workflow ==
88
89 ```
90 Contributor edits → System validates → Applied immediately → Logged → Reputation earned
91 ```
92 **No approval required** (Wikipedia model)
93 **New contributors** (<50 reputation): Limited to minor edits
94
95 == 5.5 Quality and Audit Workflow ==
96
97 {{include reference="Archive.FactHarbor 2026\.01\.20.Specification.Diagrams.Quality and Audit Workflow.WebHome"/}}
98
99 == 6. Flagging Workflow ==
100
101 ```
102 User flags issue → Categorize (abuse/quality) → Automated or manual resolution
103 ```
104 **Quality issues**: Add to improvement queue → System fix → Auto re-process
105 **Abuse**: Moderator review → Action taken
106
107 == 7. Moderation Workflow ==
108
109 **Automated pre-moderation**: 95% published automatically
110 **Moderator queue**: Only high-risk or flagged content
111 **Appeal process**: Different moderator → Governing Team if needed
112
113 == 8. System Improvement Workflow ==
114
115 **Weekly cycle**:
116 ```
117 Monday: Review error patterns
118 Tuesday-Wednesday: Develop fixes
119 Thursday: Test improvements
120 Friday: Deploy & re-process
121 Weekend: Monitor metrics
122 ```
123 **Error capture**:
124 ```
125 Error detected → Categorize → Root cause → Improvement queue → Pattern analysis
126 ```
127 **A/B Testing**:
128 ```
129 New algorithm → Split traffic (90% control, 10% test) → Run 1 week → Compare metrics → Deploy if better
130 ```
131
132 == 9. Quality Monitoring Workflow ==
133
134 **Continuous**: Every hour calculate metrics, detect anomalies
135 **Daily**: Update source track records, aggregate error patterns
136 **Weekly**: System improvement cycle, performance review
137
138 == 10. Source Track Record Workflow ==
139
140 **Initial score**: New source starts at 50 (neutral)
141 **Daily updates**: Calculate accuracy, correction frequency, update score
142 **Continuous**: All claims using source recalculated when score changes
143
144 == 11. Re-Processing Workflow ==
145
146 **Triggers**: System improvement deployed, source score updated, new evidence, error fixed
147 **Process**: Identify affected claims → Re-run AKEL → Compare → Update if better → Log change
148
149 == 12. Related Pages ==
150
151 * [[Requirements>>Archive.FactHarbor 2026\.01\.20.Specification.Requirements.WebHome]]
152 * [[Architecture>>Archive.FactHarbor 2026\.01\.20.Specification.Architecture.WebHome]]
153 * [[Data Model>>Archive.FactHarbor 2026\.01\.20.Specification.Data Model.WebHome]]