Embedding-Based Context-Aware Reranker
This addresses a bottleneck in RAG systems for applications requiring multi-passage reasoning, though it appears incremental as an enhancement to existing reranking methods.
The paper tackles the challenge of cross-passage inference in Retrieval-Augmented Generation (RAG) systems, where existing rerankers neglect issues like coreference resolution and evidence aggregation across passages. The proposed Embedding-Based Context-Aware Reranker (EBCAR) demonstrates effectiveness on the ConTEB benchmark with advantages in both accuracy and efficiency.
Retrieval-Augmented Generation (RAG) systems rely on retrieving relevant evidence from a corpus to support downstream generation. The common practice of splitting a long document into multiple shorter passages enables finer-grained and targeted information retrieval. However, it also introduces challenges when a correct retrieval would require inference across passages, such as resolving coreference, disambiguating entities, and aggregating evidence scattered across multiple sources. Many state-of-the-art (SOTA) reranking methods, despite utilizing powerful large pretrained language models with potentially high inference costs, still neglect the aforementioned challenges. Therefore, we propose Embedding-Based Context-Aware Reranker (EBCAR), a lightweight reranking framework operating directly on embeddings of retrieved passages with enhanced cross-passage understandings through the structural information of the passages and a hybrid attention mechanism, which captures both high-level interactions across documents and low-level relationships within each document. We evaluate EBCAR against SOTA rerankers on the ConTEB benchmark, demonstrating its effectiveness for information retrieval requiring cross-passage inference and its advantages in both accuracy and efficiency.