Checked other resources

• I added a very descriptive title to this issue.
• I searched the LangChain.js documentation with the integrated search.
• I used the GitHub search to find a similar question and didn't find it.
• I am sure that this is a bug in LangChain.js rather than my code.
• The bug is not resolved by updating to the latest stable version of LangChain (or the specific integration package).

Example Code

// Add a sample document with some metadata fields:
vectorStore.addDocument(new Document({ pageContent: "Page Content!", metadata: { x: "y", z: 1, w: ["a", "b", "c"] } }))

// Narrow results with one metadata field:
vectorStore.similaritySearchWithScore("test string", 5, { x:  "y" }))

// Narrow results with two metadata fields (AND):
vectorStore.similaritySearchWithScore("test string", 5, { x:  "y", z: 1 }))

// Narrow results which contains an element on a metadata field list:
vectorStore.similaritySearchWithScore("test string", 5, { w: "a" }))

Error Message and Stack Trace (if applicable)

No response

Description

This is an issue created from the following discussion:

How to use RedisVectorStoreFilterType?

Current implementation of RedisVectorStore saves the Document's metadata object as a JSON string in the metadata-key field.

This makes it impossible to search for multiple fields.

In order to make the RedisVectorStore more compatible to the Document metadata object, we need to change how metadata fields are stored and filtered.

A use case for this is a vector store of books, which metadata with fields such as author (string), year (number) and tags (list of strings) would narrow the vector search and improve its results.

We can achieve this with MemoryVectorStore filter function, and I am pretty sure we can do the same with Redis.

I can work on a PR for this if this interests the community.

Thank you.

System Info

[email protected] | MIT | deps: 17 | versions: 262