kjannette/citation_sentinel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge pull request #2 from kjannette/update-README

Browse Source 
update README.md
This commit is contained in:
S Jannette
2026-05-09 18:15:39 -04:00
committed by GitHub
parent 01a08452b7 8d98d19e7e
commit 6edee1df6e
1 changed files with 18 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
README.md
View File

@@ -42,15 +42,24 @@ This application is a RAG (Retrieval-Augmented Generation) system that allows us
When a user submits a query, the system enforces groundedness through a multi-layered strategy: When a user submits a query, the system enforces groundedness through a multi-layered strategy:
1. Retrieval constraint — The query is embedded (also via voyage-3) and compared against stored chunk vectors using cosine similarity, returning the top 20 candidates. Only user-supplied source material is searched; the system has no web search capability. 1. **Retrieval constraint** — The query is embedded (via Voyage AI voyage-3) and compared against
stored chunk vectors using cosine similarity, returning the top 20 candidates.
3. Reranking for precision — Those 20 candidates are sent to Voyage AI's rerank-2 cross-encoder, which re-scores each query-chunk pair with deeper semantic analysis. Only the top 5 survive.
2. **Reranking for precision** — Those 20 candidates are sent to Voyage AI's rerank-2 cross-encoder,
5. Prompt-level constraint — The top 5 chunks are passed to the Primary LLM (Claude claude-opus-4-6) with an explicit system instruction: "Answer the user's question using ONLY the source documents provided below." The LLM must cite sources using bracketed indices (e.g., [1], [2]) and admit when sources are insufficient. which re-scores each query-chunk pair with deeper semantic analysis. Only the top 5 survive.
7. Schema enforcement — The LLM's response is constrained to a JSON schema requiring structured fields (answer, citedSourceIndices, followUpQuestions), and any cited source indices that don't correspond to real source groups are programmatically stripped out. 3. **Prompt-level constraint** — The top 5 chunks are passed to the "Primary LLM" (Claude opus-4-6).
The LLM must cite sources using bracketed indices (e.g., [1], [2]) and admit when sources are
9. Post-generation groundedness scoring — After the answer is generated, it is split into individual sentences, each sentence is embedded via voyage-3, and each sentence embedding is compared (cosine similarity) against the vectors of the cited chunks. The raw similarities are calibrated to a 0–1 scale and averaged, producing a single groundedness score that is surfaced to the user as a visual indicator (green/gold/red). insufficient.
4. **Schema enforcement** — The LLM's response is constrained to a JSON schema requiring structured
fields (answer, citedSourceIndices, followUpQuestions). Any cited source indices that do not
correspond to real source groups are programmatically stripped out.
5. **Post-generation groundedness scoring** — The answer is split into individual sentences. Voyage
AI voyage-3 embeds each sentence, and compares it (using cosine similarity) against the vectors
of the cited chunks. The similarity is calibrated to a 0–1 scale and averaged, producing a single
groundedness score that is surfaced to the user with a visual indicator (green/yellow/red).
## Design ## Design