Case-Insensitive Similarity Search in ChromaDB for RAG Pipeline

[Adarsh-AMT]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

Hi everyone,
We're currently working on a RAG (Retrieval-Augmented Generation) pipeline using ChromaDB and GPT-4o, and we’ve run into a case sensitivity issue during similarity search.

We're using the following code to perform the search:
vector_similarity_search = chroma_initializer.max_marginal_relevance_search(
query=question,
k=k,
fetch_k=2 * k, # Ensuring fetch_k isn't too large
lambda_mult=0.6, # Adjusting lambda for better relevance balance
where_document=where_document_filter,
)

The issue is that when the query contains a lowercase word like "ahmad", it doesn’t match a capitalized version like "Ahmad" in the documents. We want the similarity check to be case-insensitive so that matches aren't missed due to casing.

Is there any built-in parameter in ChromaDB that supports case-insensitive search without affecting performance?

Motivation

.

Proposal (If applicable)

No response