Similarity matrix of documents instead of similarity based query

[emilmirzayev]

Hi, is it possible to generate a similarity matrix of the documents instead of querying k elements based on cosine similarity using langchain?

Querying I mean this:

from langchain.text_splitter import CharacterTextSplitter, RecursiveCharacterTextSplitter

save_directory = "db"


text_splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
documents = text_splitter.create_documents(multiple_abstracts,
                                           metadatas= multiple_metadata
                                          )



db = Chroma.from_documents(documents = documents,
                           embedding= embeddings,
                           persist_directory=save_directory
                          )

query = "Which document describes glasses for use with Salesforce?"
docs = db.similarity_search(query)

docs[0]

Now, instead of searching by score, i want to create a score matrix n x n, where n is the number of documents.