Retrieval Augmentation for T5 Re-ranker using External Sources

TL;DR

This paper explores how retrieval augmentation from external sources like web search and Wikipedia can significantly enhance the performance of T5-based re-rankers in both in-domain and out-of-domain scenarios.

Contribution

It introduces a method for augmenting T5 re-rankers with external retrieval, demonstrating substantial improvements in effectiveness across various re-ranking tasks.

Findings

Retrieval augmentation improves T5 re-ranker performance

Effective in both in-domain and zero-shot out-of-domain tasks

Utilizes external sources like web search and Wikipedia

Abstract

Retrieval augmentation has shown promising improvements in different tasks. However, whether such augmentation can assist a large language model based re-ranker remains unclear. We investigate how to augment T5-based re-rankers using high-quality information retrieved from two external corpora -- a commercial web search engine and Wikipedia. We empirically demonstrate how retrieval augmentation can substantially improve the effectiveness of T5-based re-rankers for both in-domain and zero-shot out-of-domain re-ranking tasks.