Korum
Korum (Structured Debates) — sub‑module under ethiKos.
Implements five core services with defined code‑names, backed by concrete debate/stance/argument models and fixed parameters.
1) Functional Services (and expected files)
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 AI 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 |
2) Backend functionalities
-
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; integrates with Smart Vote for weighted aggregation.
-
Weighted results & cohorts: Results recomputed by expertise cohorts (EkoH) and other filters; realtime push optional via Channels/Redis.
-
Quality & moderation: Report/auto‑hide thresholds; flagged content routed to shared moderation queue.
3) Database models (OLTP)
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.
4) Supporting configuration (frozen)
-
Stance scale: −3 … +3 (0 \= neutral).
-
Expert cohort quorum (display): 12 distinct experts (Ekoh threshold per domain).
-
Moderation auto‑hide: Hide an argument after 3 independent reports.
-
AI clone training batch: 128 records.
5) Frontend & navigation
-
Routes:
/debate(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.
6) Integration points
-
EkoH & Smart Vote: Stances are aggregated using EkoH reputation to compute weighted results; outcomes can feed analytics (/reports/smart‑vote).
-
Insights module: ETL ingests vote/stance facts; dashboards render trends with export limits and privacy safeguards (k‑anonymity, hashed IDs).
7) Realtime & ops
-
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.
Summary
Korum provides structured topic management, nuanced stance capture, and threaded argumentation, with weighted consensus via EkoH/Smart Vote. Its five named services are stable integration points; the production schema covers categories, topics, stances, and arguments, while advanced AI/archive features run as services without additional OLTP tables in v14. Routes and parameters are version‑locked to ensure predictable behavior across UI, API, and analytics.