Mix-of-Granularity: Optimize the Chunking Granularity for Retrieval-Augmented Generation
This addresses a domain-specific problem for RAG systems by improving retrieval efficiency and information exploitation, though it is incremental as it builds on existing RAG and Mix-of-Expert concepts.
The paper tackles the challenge of integrating diverse knowledge sources in Retrieval-Augmented Generation (RAG) systems by introducing Mix-of-Granularity (MoG) and MoG-Graph (MoGG), which dynamically optimize chunking granularity based on queries, significantly enhancing performance in downstream tasks.
Integrating information from various reference databases is a major challenge for Retrieval-Augmented Generation (RAG) systems because each knowledge source adopts a unique data structure and follows different conventions. Retrieving from multiple knowledge sources with one fixed strategy usually leads to under-exploitation of information. To mitigate this drawback, inspired by Mix-of-Expert, we introduce Mix-of-Granularity (MoG), a method that dynamically determines the optimal granularity of a knowledge source based on input queries using a router. The router is efficiently trained with a newly proposed loss function employing soft labels. We further extend MoG to MoG-Graph (MoGG), where reference documents are pre-processed as graphs, enabling the retrieval of distantly situated snippets. Experiments demonstrate that MoG and MoGG effectively predict optimal granularity levels, significantly enhancing the performance of the RAG system in downstream tasks. The code of both MoG and MoGG are released in https://github.com/ZGChung/Mix-of-Granularity.