Reason of using _to_llama_similarities as normalization technique in Elastic Search Vector Store

[Subham0793]

Because of _to_llama_similarities function, the scores are normalized and maximum score is always at 1.0

What is the reason of this normalization technique, and why has this being called by default?

The problem may arise when some query which is not related to a particular document, is used for retrieving the chunks, and still the results will contain scores starting from 1.0 , even when there is no relation of the chunk with respect to the query.

Code snippet:

return VectorStoreQueryResult( nodes=top_k_nodes, ids=top_k_ids, similarities=_to_llama_similarities(top_k_scores), )

Line no. 623 in

https://github.com/run-llama/llama_index/blob/main/llama-index-integrations/vector_stores/llama-index-vector-stores-elasticsearch/llama_index/vector_stores/elasticsearch/base.py