Skip to Content

Wiki source code of V1.0 Requirements

Last modified by Robert Schaub on 2025/12/22 13:50
Show last authors
1 = V1.0 Requirements =
2
3 **Version:** 0.9.70
4 **Phase:** Version 1.0 (Production Launch)
5 **Priority:** CRITICAL
6 **Status:** Ready for Implementation
7
8 This page specifies the requirements that MUST be implemented for FactHarbor V1.0 production launch, based on comprehensive fact-checking industry research (December 2025).
9
10
11 == Overview ==
12
13 V1.0 adds critical requirements for:
14 * **Platform Integration** (ClaimReview schema, corrections system)
15 * **Quality Assurance** (Enhanced AKEL gates, security)
16 * **Media Verification** (Image verification, evidence archiving)
17 * **Community Safety** (Contributor protection)
18 * **Continuous Improvement** (A/B testing, quality metrics)
19
20 **Total New Requirements:** 11 (FR44-FR54, NFR11-NFR13)
21 **New User Needs:** 3 (UN-26, UN-27, UN-28)
22
23
24 == Category 1: Platform Integration & Standards Compliance ==
25
26 === FR44: ClaimReview Schema Implementation ===
27
28 **Priority:** CRITICAL
29 **Fulfills:** UN-13 (Cite verdicts), UN-14 (API access), UN-26 (Search engine visibility)
30 **Phase:** V1.0
31
32 **Purpose:** Make FactHarbor analyses discoverable via Google Fact Check Explorer and other search engines.
33
34 **Specification:**
35
36 FactHarbor must generate valid ClaimReview structured data for every published analysis following Schema.org specifications.
37
38 **Required Fields:**
39
40 {{code language="json"}}
41 {
42 "@context": "https://schema.org",
43 "@type": "ClaimReview",
44 "datePublished": "YYYY-MM-DD",
45 "url": "https://factharbor.org/claims/{claim_id}",
46 "claimReviewed": "The exact claim text",
47 "author": {
48 "@type": "Organization",
49 "name": "FactHarbor",
50 "url": "https://factharbor.org"
51 },
52 "reviewRating": {
53 "@type": "Rating",
54 "ratingValue": "1-5",
55 "bestRating": "5",
56 "worstRating": "1",
57 "alternateName": "FactHarbor likelihood score"
58 },
59 "itemReviewed": {
60 "@type": "Claim",
61 "author": {
62 "@type": "Person" or "Organization",
63 "name": "Claim author if known"
64 },
65 "datePublished": "YYYY-MM-DD if known",
66 "appearance": {
67 "@type": "CreativeWork",
68 "url": "Original claim URL if from article"
69 }
70 }
71 }
72 {{/code}}
73
74 **FactHarbor-Specific Mapping:**
75
76 Rating Value Conversion:
77 * 80-100% likelihood → 5 (Highly Supported)
78 * 60-79% likelihood → 4 (Supported)
79 * 40-59% likelihood → 3 (Mixed/Uncertain)
80 * 20-39% likelihood → 2 (Questionable)
81 * 0-19% likelihood → 1 (Refuted)
82
83 **Multiple Scenarios Handling:**
84
85 If claim has multiple scenarios with different verdicts:
86 * Generate **separate ClaimReview** for each scenario
87 * Add `disambiguatingDescription` field explaining scenario context
88 * Example: "In the context of [scenario assumptions]..."
89
90 **Implementation Requirements:**
91
92 1. Auto-generate ClaimReview JSON-LD on claim publication
93 2. Embed in HTML `<head>` section of claim page
94 3. Validate against Schema.org validator before deployment
95 4. Submit sitemap to Google Search Console
96 5. Update ClaimReview when verdict changes (FR8: Time Evolution)
97 6. Handle corrections (update dateModified field)
98
99 **Acceptance Criteria:**
100
101 * ✅ Passes Google Structured Data Testing Tool
102 * ✅ Appears in Google Fact Check Explorer within 48 hours of publication
103 * ✅ Valid JSON-LD syntax (no errors)
104 * ✅ All required fields populated
105 * ✅ Handles multi-scenario claims correctly
106 * ✅ Updates automatically on verdict changes
107
108 **Integration Points:**
109
110 * FR7: Automated Verdicts (source of rating data)
111 * FR8: Time Evolution (triggers schema updates)
112 * FR11: Audit Trail (log schema generation/updates)
113 * FR45: Corrections (updates schema on corrections)
114
115 **Resources:**
116
117 * ClaimReview Project: https://www.claimreviewproject.com
118 * Schema.org ClaimReview: https://schema.org/ClaimReview
119 * Google Fact Check Guidelines: https://developers.google.com/search/docs/appearance/fact-check
120
121
122 === FR45: User Corrections Notification System ===
123
124 **Priority:** CRITICAL
125 **Fulfills:** IFCN Principle 5 (Open & Honest Corrections), EFCSN compliance
126 **Phase:** V1.0
127
128 **Purpose:** When claim analyses are corrected, notify users who previously viewed the claim.
129
130 **Specification:**
131
132 **Correction Types:**
133
134 1. **Major Correction:** Verdict changes category (e.g., "Supported" → "Refuted")
135 2. **Significant Correction:** Likelihood score changes >20%
136 3. **Minor Correction:** Evidence additions, source quality updates
137 4. **Scenario Addition:** New scenario added to existing claim
138
139 **Notification Mechanisms:**
140
141 **1. In-Page Banner (Required):**
142
143 {{code}}
144 [!] CORRECTION NOTICE
145 This analysis was updated on [DATE]. [View what changed] [Dismiss]
146
147 Major changes:
148 • Verdict changed from "Likely True (75%)" to "Uncertain (45%)"
149 • New contradicting evidence added from [Source]
150 • Scenario 2 updated with additional context
151
152 [See full correction log]
153 {{/code}}
154
155 **Display Rules:**
156 * Show banner on ALL pages displaying the claim
157 * Banner persists for 30 days after correction
158 * "Corrections" count badge on claim card
159 * Timestamp on every verdict: "Last updated: [datetime]"
160
161 **2. Correction Log Page (Required):**
162
163 * Public changelog at `/claims/{id}/corrections`
164 * Displays:
165 * Date/time of correction
166 * What changed (before/after comparison)
167 * Why changed (reason if provided)
168 * Who made change (AKEL auto-update vs. contributor override)
169 * Diff view of changes
170
171 **3. Email Notifications (Optional for users):**
172
173 * Send to users who bookmarked/shared claim
174 * Subject: "FactHarbor Correction: [Claim title]"
175 * Include summary of changes
176 * Link to updated analysis
177 * Unsubscribe option
178
179 **4. RSS/API Feed (Required):**
180
181 * Corrections feed at `/corrections.rss`
182 * API endpoint: `GET /api/corrections?since={timestamp}`
183 * Enables external monitoring
184 * Machine-readable format
185
186 **IFCN Compliance Requirements:**
187
188 * Corrections policy published at `/corrections-policy`
189 * User can report suspected errors via `/report-error/{claim_id}`
190 * Link to IFCN complaint process (if FactHarbor becomes signatory)
191 * Scrupulous transparency: **never** silently edit
192 * All corrections permanent and public
193
194 **Acceptance Criteria:**
195
196 * ✅ Banner appears within 60 seconds of correction
197 * ✅ Correction log is permanent and public
198 * ✅ Email notifications deliver <5 minutes
199 * ✅ RSS feed updates in real-time
200 * ✅ Mobile-responsive banner design
201 * ✅ Accessible (screen reader compatible)
202 * ✅ Cannot be dismissed permanently (reappears for 30 days)
203
204 **Integration Points:**
205
206 * FR8: Time Evolution (triggers corrections)
207 * FR11: Audit Trail (source of correction data)
208 * NFR3: Transparency (public correction log)
209 * FR44: ClaimReview (updates schema)
210
211
212 == Category 2: Quality Assurance ==
213
214 === NFR11: AKEL Quality Assurance Framework ===
215
216 **Priority:** CRITICAL
217 **Fulfills:** AI safety, IFCN methodology transparency
218 **Phase:** V1.0
219
220 **Purpose:** Prevent AI hallucinations and low-quality outputs through multi-layer automated checks.
221
222 **Specification:**
223
224 This enhances the existing 4 quality gates with more detailed specifications and confidence thresholds.
225
226 **Gate 1: Claim Extraction Validation**
227
228 **Purpose:** Ensure extracted claims are factual assertions (not opinions/predictions)
229
230 **Automated Checks:**
231
232 1. **Factual Statement Test:** Is this verifiable? (Yes/No)
233 2. **Opinion Detection:** Contains hedging language? ("I think", "probably", "might")
234 3. **Future Prediction Test:** Makes claim about future events?
235 4. **Specificity Score:** Contains specific entities, numbers, dates?
236
237 **Thresholds:**
238
239 * Factual: Must be "Yes"
240 * Opinion markers: <2 hedging phrases
241 * Specificity: ≥3 specific elements
242
243 **Action if Failed:**
244
245 * Flag claim as "Non-verifiable" or "Opinion"
246 * Do NOT generate verdict
247 * Display to user: "This appears to be an opinion rather than a factual claim"
248 * Log pattern for system improvement
249
250
251 **Gate 2: Evidence Relevance Validation**
252
253 **Purpose:** Ensure AI-linked evidence actually relates to claim
254
255 **Automated Checks:**
256
257 1. **Semantic Similarity Score:** Evidence text vs. claim (using embeddings)
258 2. **Entity Overlap:** Do evidence and claim mention same people/places/things?
259 3. **Contradiction Detection:** Does evidence discuss the claim topic?
260
261 **Thresholds:**
262
263 * Similarity: ≥0.6 (cosine similarity)
264 * Entity overlap: ≥1 shared entity
265 * Topic relevance: ≥0.5
266
267 **Action if Failed:**
268
269 * Discard irrelevant evidence
270 * If <2 relevant evidence items remain, verdict = "Insufficient Evidence"
271 * Block publication if below threshold
272 * Log pattern for search improvement
273
274
275 **Gate 3: Scenario Coherence Check**
276
277 **Purpose:** Validate scenario assumptions are logical and complete
278
279 **Automated Checks:**
280
281 1. **Completeness:** Scenario has all required fields
282 2. **Internal Consistency:** Assumptions don't contradict each other
283 3. **Distinguishability:** Scenarios are meaningfully different (not duplicates)
284
285 **Thresholds:**
286
287 * Required fields: 100% populated
288 * Contradiction score: <0.3 (self-contradiction embedding)
289 * Scenario similarity: <0.8 (between scenarios)
290
291 **Action if Failed:**
292
293 * Merge duplicate scenarios
294 * Flag inconsistent assumptions for sampling audit
295 * Reduce confidence score by 20%
296
297
298 **Gate 4: Verdict Confidence Assessment**
299
300 **Purpose:** Only publish high-confidence verdicts
301
302 **Automated Checks:**
303
304 1. **Evidence Count:** Minimum 2 sources (EFCSN standard)
305 2. **Source Quality:** Average source reliability ≥0.6
306 3. **Evidence Agreement:** Supporting vs. contradicting ratio
307 4. **Uncertainty Factors:** Number of explicit uncertainties
308
309 **Confidence Tiers:**
310
311 {{code}}
312 HIGH (80-100%):
313 - ≥3 high-quality sources
314 - >80% agreement
315 - <2 uncertainty factors
316 - Publish immediately (all risk tiers)
317
318 MEDIUM (50-79%):
319 - 2-3 sources
320 - 60-80% agreement
321 - 2-4 uncertainty factors
322 - Publish with standard labels
323
324 LOW (0-49%):
325 - <2 sources OR
326 - <60% agreement OR
327 - >4 uncertainty factors
328 - BLOCK publication
329 {{/code}}
330
331 **Publication Rules:**
332
333 * HIGH confidence: Publish immediately
334 * MEDIUM confidence: Publish with "May contain uncertainties" label
335 * LOW confidence: Block, improve system
336
337 **Acceptance Criteria:**
338
339 * ✅ All 4 gates implemented
340 * ✅ Thresholds configurable (for A/B testing)
341 * ✅ Gate failures logged with details
342 * ✅ Confidence scores accurate (validated through sampling audits)
343 * ✅ <5% hallucination rate (measured via audits)
344
345 **Integration Points:**
346
347 * FR7: Automated Verdicts (applies gates)
348 * AKEL: Quality gates enforcement
349 * NFR13: Quality metrics (reports gate performance)
350
351
352 === NFR12: Advanced Security Controls ===
353
354 **Priority:** CRITICAL
355 **Fulfills:** Production security requirements
356 **Phase:** V1.0
357
358 **Specification:**
359
360 **Essential Security (V1.0 Launch):**
361
362 1. **DDoS Protection:**
363 * Rate limiting: 100 requests/hour per IP (content submission)
364 * Cloudflare or equivalent
365 * Automatic IP blocking on abuse
366
367 2. **API Rate Limiting:**
368 * 1000 requests/hour per API key
369 * Burst allowance: 50 requests/minute
370 * 429 responses with retry-after headers
371
372 3. **Audit Logging:**
373 * All moderation actions logged
374 * All system changes logged
375 * Logs retained 2 years
376 * Tamper-proof logging
377
378 4. **Input Validation:**
379 * Sanitize all user inputs
380 * Prevent SQL injection
381 * Prevent XSS attacks
382 * Max input sizes enforced
383
384 5. **Authentication:**
385 * OAuth 2.0 for API access
386 * Secure session management
387 * Password requirements (if applicable)
388
389 **Full Security (V1.1+):**
390
391 6. **Penetration Testing:** Annual third-party tests
392 7. **Vulnerability Scanning:** Automated weekly scans
393 8. **Security Incident Response:** Documented procedures
394 9. **Data Encryption:** At rest and in transit
395 10. **Access Control:** Role-based permissions
396
397 **Acceptance Criteria:**
398
399 * ✅ Rate limits enforced
400 * ✅ DDoS protection active
401 * ✅ All logs captured
402 * ✅ Input validation prevents common attacks
403 * ✅ API authentication required
404
405
406 === NFR13: Reference-Free Quality Metrics ===
407
408 **Priority:** CRITICAL (POC2 onwards)
409 **Fulfills:** NFR3 (Transparency), continuous improvement monitoring
410 **Phase:** POC2, Beta 0, V1.0
411
412 **Purpose:** Measure AKEL quality without requiring human-labeled ground truth.
413
414 **Specification:**
415
416 **Metrics Dashboard (Public):**
417
418 **1. Consistency Metrics:**
419
420 * **Cross-Source Consistency:** Do verdicts align with evidence?
421 * **Temporal Consistency:** Do verdicts remain stable over time (when evidence unchanged)?
422 * **Scenario Consistency:** Do related scenarios have coherent verdicts?
423
424 **2. Completeness Metrics:**
425
426 * **Evidence Retrieval Rate:** % of claims with ≥2 sources
427 * **Contradiction Search Coverage:** % of claims with counter-evidence searched
428 * **Source Diversity:** Number of distinct sources per claim
429
430 **3. Confidence Calibration:**
431
432 * **Confidence vs. Evidence Strength:** Does high confidence correlate with strong evidence?
433 * **Confidence Distribution:** Are verdicts appropriately uncertain?
434
435 **4. Quality Gate Performance:**
436
437 * **Gate Pass Rates:** % passing each gate
438 * **Gate Failure Reasons:** What causes most failures?
439 * **Gate Effectiveness:** Sampling audit validation of gates
440
441 **5. User Engagement Metrics:**
442
443 * **Correction Rate:** How often do users flag issues?
444 * **Appeal Rate:** How often are corrections requested?
445 * **User Satisfaction:** Survey results
446
447 **Dashboard Features:**
448
449 * Public access at `/quality-metrics`
450 * Updated daily
451 * Historical trends (30/90/365 days)
452 * Breakdown by risk tier
453 * Download raw data (CSV)
454
455 **Acceptance Criteria:**
456
457 * ✅ Dashboard publicly accessible
458 * ✅ Updates daily
459 * ✅ All metrics implemented
460 * ✅ Historical data retained
461 * ✅ Transparent methodology explained
462
463
464 == Category 3: Media Verification ==
465
466 === FR46: Image Verification System ===
467
468 **Priority:** CRITICAL
469 **Fulfills:** UN-27 (Visual claim verification)
470 **Phase:** V1.0 (Basic), V1.1 (Extended)
471
472 **Purpose:** Enable users to verify image-based claims.
473
474 **V1.0 Specification (Basic):**
475
476 **1. Reverse Image Search:**
477
478 * Integration with Google Image Search API
479 * Integration with TinEye API
480 * Display earliest known appearance
481 * Show similar/modified versions
482
483 **2. Metadata Analysis:**
484
485 * EXIF data extraction
486 * Creation date/time
487 * Camera/device information
488 * GPS location (if available)
489 * Edit history (if available)
490
491 **3. Basic Manipulation Detection:**
492
493 * Error Level Analysis (ELA)
494 * Flag obviously manipulated images
495 * Not AI-powered detection (V1.1+)
496
497 **UI Workflow:**
498
499 {{code}}
500 User uploads image
501 ↓
502 System runs reverse search
503 ↓
504 System extracts metadata
505 ↓
506 System performs ELA
507 ↓
508 Results displayed:
509 - Earliest known appearance
510 - Similar images found
511 - Metadata (camera, date, location)
512 - Manipulation indicators (if any)
513 ↓
514 User can create claim based on findings
515 {{/code}}
516
517 **V1.1 Specification (Extended - Future):**
518
519 * AI-powered deepfake detection
520 * Acoustic signature analysis (for videos)
521 * Advanced forensic tools (noise patterns, compression artifacts)
522
523 **Acceptance Criteria (V1.0):**
524
525 * ✅ Reverse search functional (Google + TinEye)
526 * ✅ Metadata extracted correctly
527 * ✅ ELA results displayed
528 * ✅ User-friendly interface
529 * ✅ Results help users make informed decisions
530
531
532 === FR47: Archive.org Integration ===
533
534 **Priority:** CRITICAL
535 **Fulfills:** Evidence persistence, FR5 (Evidence linking)
536 **Phase:** V1.0
537
538 **Purpose:** Ensure evidence remains accessible even if original sources are deleted.
539
540 **Specification:**
541
542 **Automatic Archiving:**
543
544 When AKEL links evidence:
545 1. Check if URL already archived (Wayback Machine API)
546 2. If not, submit for archiving (Save Page Now API)
547 3. Store both original URL and archive URL
548 4. Display both to users
549
550 **Archive Display:**
551
552 {{code}}
553 Evidence Source: [Original URL]
554 Archived: [Archive.org URL] (Captured: [date])
555
556 [View Original] [View Archive]
557 {{/code}}
558
559 **Fallback Logic:**
560
561 * If original URL unavailable → Auto-redirect to archive
562 * If archive unavailable → Display warning
563 * If both unavailable → Flag for manual review
564
565 **API Integration:**
566
567 * Use Wayback Machine Availability API
568 * Use Save Page Now API (SPNv2)
569 * Rate limiting: 15 requests/minute (Wayback limit)
570
571 **Acceptance Criteria:**
572
573 * ✅ All evidence URLs auto-archived
574 * ✅ Archive links displayed to users
575 * ✅ Fallback to archive if original unavailable
576 * ✅ API rate limits respected
577 * ✅ Archive status visible in evidence display
578
579
580 == Category 4: Community Safety ==
581
582 === FR48: Contributor Safety Framework ===
583
584 **Priority:** CRITICAL
585 **Fulfills:** UN-28 (Safe contribution environment)
586 **Phase:** V1.0
587
588 **Purpose:** Protect contributors from harassment, doxxing, and coordinated attacks.
589
590 **Specification:**
591
592 **1. Privacy Protection:**
593
594 * **Optional Pseudonymity:** Contributors can use pseudonyms
595 * **Email Privacy:** Emails never displayed publicly
596 * **Profile Privacy:** Contributors control what's public
597 * **IP Logging:** Only for abuse prevention, not public
598
599 **2. Harassment Prevention:**
600
601 * **Automated Toxicity Detection:** Flag abusive comments
602 * **Personal Information Detection:** Auto-block doxxing attempts
603 * **Coordinated Attack Detection:** Identify brigading patterns
604 * **Rapid Response:** Moderator alerts for harassment
605
606 **3. Safety Features:**
607
608 * **Block Users:** Contributors can block harassers
609 * **Private Contributions:** Option to contribute anonymously
610 * **Report Harassment:** One-click harassment reporting
611 * **Safety Resources:** Links to support resources
612
613 **4. Moderator Tools:**
614
615 * **Quick Ban:** Immediately block abusers
616 * **Pattern Detection:** Identify coordinated attacks
617 * **Appeal Process:** Fair review of moderation actions
618 * **Escalation:** Serious threats escalated to authorities
619
620 **5. Trusted Contributor Protection:**
621
622 * **Enhanced Privacy:** Additional protection for high-profile contributors
623 * **Verification:** Optional identity verification (not public)
624 * **Legal Support:** Resources for contributors facing legal threats
625
626 **Acceptance Criteria:**
627
628 * ✅ Pseudonyms supported
629 * ✅ Toxicity detection active
630 * ✅ Doxxing auto-blocked
631 * ✅ Harassment reporting functional
632 * ✅ Moderator tools implemented
633 * ✅ Safety policy published
634
635
636 == Category 5: Continuous Improvement ==
637
638 === FR49: A/B Testing Framework ===
639
640 **Priority:** CRITICAL
641 **Fulfills:** Continuous system improvement
642 **Phase:** V1.0
643
644 **Purpose:** Test and measure improvements to AKEL prompts, algorithms, and workflows.
645
646 **Specification:**
647
648 **Test Capabilities:**
649
650 1. **Prompt Variations:**
651 * Test different claim extraction prompts
652 * Test different verdict generation prompts
653 * Measure: Accuracy, clarity, completeness
654
655 2. **Algorithm Variations:**
656 * Test different source scoring algorithms
657 * Test different confidence calculations
658 * Measure: Audit accuracy, user satisfaction
659
660 3. **Workflow Variations:**
661 * Test different quality gate thresholds
662 * Test different risk tier assignments
663 * Measure: Publication rate, quality scores
664
665 **Implementation:**
666
667 * **Traffic Split:** 50/50 or 90/10 splits
668 * **Randomization:** Consistent per claim (not per user)
669 * **Metrics Collection:** Automatic for all variants
670 * **Statistical Significance:** Minimum sample size calculation
671 * **Rollout:** Winner promoted to 100% traffic
672
673 **A/B Test Workflow:**
674
675 {{code}}
676 1. Hypothesis: "New prompt improves claim extraction"
677 2. Design test: Control vs. Variant
678 3. Define metrics: Extraction accuracy, completeness
679 4. Run test: 7-14 days, minimum 100 claims each
680 5. Analyze results: Statistical significance?
681 6. Decision: Deploy winner or iterate
682 {{/code}}
683
684 **Acceptance Criteria:**
685
686 * ✅ A/B testing framework implemented
687 * ✅ Can test prompt variations
688 * ✅ Can test algorithm variations
689 * ✅ Metrics automatically collected
690 * ✅ Statistical significance calculated
691 * ✅ Results inform system improvements
692
693
694 === FR54: Evidence Deduplication ===
695
696 **Priority:** CRITICAL (POC2/Beta)
697 **Fulfills:** Accurate evidence counting, quality metrics
698 **Phase:** POC2, Beta 0, V1.0
699
700 **Purpose:** Avoid counting the same source multiple times when it appears in different forms.
701
702 **Specification:**
703
704 **Deduplication Logic:**
705
706 1. **URL Normalization:**
707 * Remove tracking parameters (?utm_source=...)
708 * Normalize http/https
709 * Normalize www/non-www
710 * Handle redirects
711
712 2. **Content Similarity:**
713 * If two sources have >90% text similarity → Same source
714 * If one is subset of other → Same source
715 * Use fuzzy matching for minor differences
716
717 3. **Cross-Domain Syndication:**
718 * Detect wire service content (AP, Reuters)
719 * Mark as single source if syndicated
720 * Count original publication only
721
722 **Display:**
723
724 {{code}}
725 Evidence Sources (3 unique, 5 total):
726
727 1. Original Article (NYTimes)
728 - Also appeared in: WashPost, Guardian (syndicated)
729
730 2. Research Paper (Nature)
731
732 3. Official Statement (WHO)
733 {{/code}}
734
735 **Acceptance Criteria:**
736
737 * ✅ URL normalization works
738 * ✅ Content similarity detected
739 * ✅ Syndicated content identified
740 * ✅ Unique vs. total counts accurate
741 * ✅ Improves evidence quality metrics
742
743
744 == Additional Requirements (Lower Priority) ==
745
746 === FR50: OSINT Toolkit Integration ===
747
748 **Priority:** HIGH (V1.1)
749 **Fulfills:** Advanced media verification
750 **Phase:** V1.1
751
752 **Purpose:** Integrate open-source intelligence tools for advanced verification.
753
754 **Tools to Integrate:**
755 * InVID/WeVerify (video verification)
756 * Bellingcat toolkit
757 * Additional TBD based on V1.0 learnings
758
759
760 === FR51: Video Verification System ===
761
762 **Priority:** HIGH (V1.1)
763 **Fulfills:** UN-27 (Visual claims), advanced media verification
764 **Phase:** V1.1
765
766 **Purpose:** Verify video-based claims.
767
768 **Specification:**
769 * Keyframe extraction
770 * Reverse video search
771 * Deepfake detection (AI-powered)
772 * Metadata analysis
773 * Acoustic signature analysis
774
775
776 === FR52: Interactive Detection Training ===
777
778 **Priority:** MEDIUM (V1.5)
779 **Fulfills:** Media literacy education
780 **Phase:** V1.5
781
782 **Purpose:** Teach users to identify misinformation.
783
784 **Specification:**
785 * Interactive tutorials
786 * Practice exercises
787 * Detection quizzes
788 * Gamification elements
789
790
791 === FR53: Cross-Organizational Sharing ===
792
793 **Priority:** MEDIUM (V1.5)
794 **Fulfills:** Collaboration with other fact-checkers
795 **Phase:** V1.5
796
797 **Purpose:** Share findings with IFCN/EFCSN members.
798
799 **Specification:**
800 * API for fact-checking organizations
801 * Structured data exchange
802 * Privacy controls
803 * Attribution requirements
804
805
806 == Summary ==
807
808 **V1.0 Critical Requirements (Must Have):**
809
810 * FR44: ClaimReview Schema ✅
811 * FR45: Corrections Notification ✅
812 * FR46: Image Verification ✅
813 * FR47: Archive.org Integration ✅
814 * FR48: Contributor Safety ✅
815 * FR49: A/B Testing ✅
816 * FR54: Evidence Deduplication ✅
817 * NFR11: Quality Assurance Framework ✅
818 * NFR12: Security Controls ✅
819 * NFR13: Quality Metrics Dashboard ✅
820
821 **V1.1+ (Future):**
822
823 * FR50: OSINT Integration
824 * FR51: Video Verification
825 * FR52: Detection Training
826 * FR53: Cross-Org Sharing
827
828
829 **Total:** 11 critical requirements for V1.0