Bible Passage Finder
Menu

How it works

Retrieval first, explanation second

Bible Passage Finder is designed to find real scripture references from remembered lines, lyric fragments, quotes, and paraphrases.

Retrieval-first search pipelineFour connected stages show a user query moving through embedding, the local WEB-C index, hybrid scoring, and verified passage output.Query embeddingbounded evidenceWEB-C indexbounded evidenceHybrid scorebounded evidenceVerifiedreal passage
Generation never gets to guess. The search flow keeps references inside the retrieved candidate set.
1

Embed the query

The user’s phrase is converted into a semantic vector through OpenRouter.

2

Search the local Bible index

The query is compared against prebuilt WEB-C verse, short-passage, and context chunks.

3

Blend semantic and lexical signals

Exact wording, keyword overlap, and semantic similarity are combined into a hybrid score.

4

Group and rerank candidates

Overlapping passages are grouped, then ambiguous candidates can be reranked by a cross-encoder model.

Frequently asked questions

Can Bible Passage Finder identify paraphrased verses?

Yes. It uses semantic search across verse, short-passage, and context chunks, so it can find passages when the wording is similar but not exact.

Does the app make up Bible references?

No. The system retrieves candidates from the WEB-C Bible index first. Reranking can only reorder candidates that already exist in the retrieved scripture data.

Which Bible translation does it use?

The current Bible Passage Finder uses the World English Bible Catholic Edition from eBible.org, internally identified as WEB-C.