Preliminary Ranking-Based Selection for Optimized Retriever Configuration in RAG Systems

Retrieval Augmented Generation (RAG) systems rely on efficient retrievers to fetch relevant documents, with their performance influenced by factors such as chunking methods and embedding models. These components determine how documents are segmented and semantically represented, directly impacting retrieval effectiveness. To enhance retriever performance, this study explores the construction of an optimizer capable of selecting the best configuration for document retrieval within a predefined solution space. The selection of technologies is a critical step, considering project constraints, query nature, and domain-specific requirements. A key challenge is filtering out unsuitable technologies while ensuring optimal performance.