Hello everyone,
I would like to get to know about the similarity search in Neo4j and how it works. Suppose that we have different nodes and each node has indexed vector embeddings. When we would like to find the most similar nodes to the query, the most similar ones are found by HNSW (Hierarchical Navigable Small World). It creates different layers with different nodes, and find the most optimized layer and look for the answer on that layer, which is a kind of ANN algorithm.
What I would like to know here is that if we might miss some important nodes in that case.
Do you think it makes the chatbot less creative by giving the same nodes as a result for similar questions?
How can we approach it if we would like to catch all of the nodes similar to the query?