EIEC
Ethnicity and (In)Equality in Early Christianity
How it works
From query to grounded answer
1. Query understanding
The system converts the user query into a semantic embedding so it can match meaning rather than only exact words.
2. Corpus retrieval
The query embedding is compared with the corpus using a vector search. The top passages are retrieved by semantic similarity.
3. Passage reranking
Retrieved passages are reranked with a relevance model to surface the most scholarly and contextually useful results.
4. Evidence-based answer generation
The final answer is generated from the top passages only. This keeps the response grounded in the source material and reduces hallucination.