Schema normalisation
source files use different column names, scales, and missing-value patterns.
AI RAG over restaurant reviews. Ask a question, inspect the evidence, open the demo.
Evidence boundary
Data boundary
A documented proof-of-concept run over four heterogeneous review inputs. The repository records 1,660 indexed review records; it is not a production client corpus.
Evidence type
repository fact
PoC dataset and run; no production SLA or client outcome claimed.
Heterogeneous review files need one queryable, source-linked interface without hiding the underlying evidence.
Follow review inputs through normalisation, Titan embeddings, OpenSearch retrieval, and a source-linked answer.
Separate review files with inconsistent schemas and manual inspection.
One inspectable PoC flow from normalisation to source-linked answers.
The PoC demonstrates a complete path from review ingestion to a source-linked answer in a browser demo.
Every signal is classified as a repository fact, recorded test, configuration, or disclosed data boundary.
source files use different column names, scales, and missing-value patterns.
generated answers must retain a path to supporting review records.
the browser, API, agent, Lambda functions, and index must share a stable contract.
The active PoC uses Titan embeddings, OpenSearch KNN retrieval, a Bedrock Agent, Lambda functions, and API Gateway.