Perform Hybrid Search on Legal Datasets Using Qdrant and BM25/mxbai

The workflow is designed to evaluate the effectiveness of hybrid search techniques on a legal dataset. It utilizes Qdrant for managing collections and performing searches, leveraging both BM25 for keyword-based retrieval and mxbai for semantic search. The process includes retrieving dataset splits, filtering relevant questions, performing searches, and evaluating the results using Reciprocal Rank Fusion (RRF). The final output provides a quality estimate of the hybrid retrieval setup, which can be further refined for building a legal RAG chatbot.