Korum (Structured Debates) — sub-module under ethiKos.
Owns debate artifacts and stance facts (−3…+3). Smart Vote publishes outcomes as baseline + declared readings; EkoH may be referenced by lenses via snapshot ids.
Each service name is stable and maps to a dedicated Django service module (file paths reflect the cookiecutter layout; exact filenames may vary by repo).
| Display name | Code name / service | Purpose / behavior | Likely file or module |
|---|---|---|---|
| Structured Debates | structured_debate | Create and manage ordered debate sequences and topics. | ethikos/services/structured_debate.py |
| Klônes IA | ai_clone_management | Manage expert-emulating agents for continuity/testing. | ethikos/services/ai_clone_management.py |
| Comparative Analysis | comparative_argument_analysis | Compare arguments to surface convergences/divergences. | ethikos/services/comparative_argument_analysis.py |
| Public Archiving | public_debate_archive | Produce immutable debate snapshots for transparency. | ethikos/services/public_debate_archive.py |
| Automated Summaries | automated_debate_summary | Generate concise, structured debate outcome digests. | ethikos/services/automated_debate_summary.py |
Debate lifecycle: CRUD for topics with status transitions (open/closed/archived), category assignment, and owner/moderation rules; exposed via DRF.
Threaded arguments: Nested replies under each topic with optional “pro/con” flag and moderation hooks.
Nuanced stance capture: Integer stance scale −3…+3 per user/topic; stance events are the canonical upstream facts for deliberation.
Readings & cohorts (derived): Outcomes are computed/published by Smart Vote using optional declared lenses (cohorts, roles, locality, time windows). If a lens uses EkoH, it MUST bind to an EkoH snapshot id (domain weights + ethics multipliers + eligibility thresholds).
Quality & moderation: Report/auto-hide thresholds; flagged content routed to shared moderation queue.
Actual implemented tables for Korum (planned AI/summary/archive tables are intentionally omitted in v14).
| Table / Model | Purpose | Key fields |
|---|---|---|
EthikosCategory | Thematic categories for debates. | id, name, description |
EthikosTopic | Debate topic/question. | id, title, status, start_date, end_date |
EthikosStance | User stance on a topic (−3…+3). | id, topic (FK), user (FK), value |
EthikosArgument | User argument/post (threaded). | id, topic (FK), author (FK), content, parent (FK), side (enum, optional) |
Note: AI clones, comparative-analysis logs, public archives, and debate summaries are listed as services but not present as tables in the current schema snapshot.
Stance scale: −3 … +3 (0 = neutral).
Expert cohort quorum (display): 12 distinct experts (eligibility derived from EkoH domain thresholds for the relevant topic tags).
Moderation auto-hide: Hide an argument after 3 independent reports.
AI clone training batch: 128 records.
Routes: /ethikos/korum (Debate Hub: Open / Archived / Start New), /ethikos/insights (opinion analytics dashboards).
Behavior: Stance slider (−3…+3), live tallies, cohort filters, threaded arguments; analytics readouts live under Insights.
EkoH + Smart Vote:
ekoh_snapshot_id).Insights module: ETL ingests stance and vote facts; dashboards render trends with export limits and privacy safeguards (k-anonymity, hashed IDs).
Push updates: Optional WebSocket broadcasts via Django Channels + Redis for stance/result changes.
Caching & rate control: Use Redis caching for common cohort filters; apply API throttles consistent with platform policy.
Korum provides structured topic management, nuanced stance capture, and threaded argumentation. Korum owns the upstream facts; Smart Vote publishes baseline + declared readings (derived outputs), optionally bound to EkoH snapshot ids for auditability.