Tony_tech/athena-oracle
3
0
Fork 0
Code Issues 6 Pull Requests 2 Actions Packages Projects Releases Wiki Activity

No semantic/embedding relationship layer #6

New Issue
Open
opened 2026-07-10 15:39:26 +00:00 by Leonard · 0 comments
Leonard commented 2026-07-10 15:39:26 +00:00
Copy Link
Copy Source

Category: Roadmap · Priority: MED

Summary
theme_scan is keyword regex over 4 hardcoded themes (tool-call, context, compute, trust). Cross-source linkage is blind. nomic-embed-text (768-dim, CPU-friendly) would replace regex themes with cosine clustering and enable semantic dedup + cross-source relationship. Not built.

Why it matters
Keyword matching misses paraphrased/adjacent topics. Dedup is exact-only (source+source_id), so rephrased repeats slip through — the exact "source-expansion ≠ real work" failure mode.

What happens if not done
Weaker trend detection; semantic near-duplicates inflate signal counts; cross-source synthesis unavailable.

Suggested approach
Add embeddings table (768 floats/entry, ~3KB each). Embed on ingest via Ollama nomic-embed-text. Replace regex themes with per-cycle cosine clustering. First real LLM dependency — pairs with issue #8 and #10 (containerize).

**Category:** Roadmap · **Priority:** MED **Summary** `theme_scan` is keyword regex over 4 hardcoded themes (tool-call, context, compute, trust). Cross-source linkage is blind. nomic-embed-text (768-dim, CPU-friendly) would replace regex themes with cosine clustering and enable semantic dedup + cross-source relationship. Not built. **Why it matters** Keyword matching misses paraphrased/adjacent topics. Dedup is exact-only (`source+source_id`), so rephrased repeats slip through — the exact "source-expansion ≠ real work" failure mode. **What happens if not done** Weaker trend detection; semantic near-duplicates inflate signal counts; cross-source synthesis unavailable. **Suggested approach** Add `embeddings` table (768 floats/entry, ~3KB each). Embed on ingest via Ollama nomic-embed-text. Replace regex themes with per-cycle cosine clustering. First real LLM dependency — pairs with issue #8 and #10 (containerize).
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
No items
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
HA_Researcher Ty tonyjbala (Tony Balascio)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Tony_tech/athena-oracle#6
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?