Implement evidence quality grading and contradiction detection system #12

Closed
opened 2026-04-03 18:17:59 +00:00 by Mark · 1 comment
Owner

Problem

Current system has no way to detect contradictory findings or resolve conflicts in data.

Requirements

Evidence Quality System

  • Use Admiralty source grading (A-F)
  • Use Admiralty information grading (1-6)
  • Track confidence levels per entity
  • Flag low-confidence findings

Contradiction Detection

  • Compare new findings against existing
  • Flag conflicting data
  • Track provenance chains
  • Resolution workflow for conflicts

Confidence Aggregation

  • Aggregate confidence from multiple sources
  • Calculate weighted confidence scores
  • Track evidence supporting each finding
## Problem Current system has no way to detect contradictory findings or resolve conflicts in data. ## Requirements ### Evidence Quality System - Use Admiralty source grading (A-F) - Use Admiralty information grading (1-6) - Track confidence levels per entity - Flag low-confidence findings ### Contradiction Detection - Compare new findings against existing - Flag conflicting data - Track provenance chains - Resolution workflow for conflicts ### Confidence Aggregation - Aggregate confidence from multiple sources - Calculate weighted confidence scores - Track evidence supporting each finding
Author
Owner

Research Phase Complete — Closing Issue #12

Core system implemented in commit d9c568c (fixes #23).

✅ Implemented

  • Admiralty source grading (A-F) based on module source type
  • Admiralty info grading (1-6) based on confidence level
  • Content hashing for evidence deduplication
  • Contradiction detection (creates Finding records with category="contradiction")
  • Provenance tracking via supporting_evidence_ids
  • Confidence aggregation and weighted scoring via AggregatedIntelligence

📋 Remaining (follow-up items)

  1. Conflict resolution workflow — UI/workflow to review & resolve detected contradictions
  2. Explicit low-confidence flagging — automated flagging of evidence below confidence threshold

Closing as the core implementation is complete. Follow-up issues can be filed for the enhancement items.

## Research Phase Complete — Closing Issue #12 **Core system implemented** in commit `d9c568c` (fixes #23). ### ✅ Implemented - Admiralty source grading (A-F) based on module source type - Admiralty info grading (1-6) based on confidence level - Content hashing for evidence deduplication - Contradiction detection (creates Finding records with category="contradiction") - Provenance tracking via supporting_evidence_ids - Confidence aggregation and weighted scoring via AggregatedIntelligence ### 📋 Remaining (follow-up items) 1. **Conflict resolution workflow** — UI/workflow to review & resolve detected contradictions 2. **Explicit low-confidence flagging** — automated flagging of evidence below confidence threshold Closing as the core implementation is complete. Follow-up issues can be filed for the enhancement items.
Mark 2026-06-08 22:32:56 +00:00
Sign in to join this conversation.
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference
Mark/osint-hub#12
No description provided.