GNN-encoder: Learning a Dual-encoder Architecture via Graph Neural Networks for Dense Passage Retrieval

TL;DR

This paper introduces a GNN-encoder that enhances dual-encoder dense passage retrieval by incorporating interaction information through graph neural networks, achieving state-of-the-art results while maintaining efficiency.

Contribution

The paper proposes a novel GNN-encoder architecture that fuses query and passage information via graph neural networks, improving accuracy without sacrificing efficiency in dense retrieval.

Findings

Significantly outperforms existing models on MSMARCO, Natural Questions, and TriviaQA datasets.

Achieves new state-of-the-art performance on these datasets.

Maintains the efficiency of dual-encoder models while enhancing their effectiveness.

Abstract

Recently, retrieval models based on dense representations are dominant in passage retrieval tasks, due to their outstanding ability in terms of capturing semantics of input text compared to the traditional sparse vector space models. A common practice of dense retrieval models is to exploit a dual-encoder architecture to represent a query and a passage independently. Though efficient, such a structure loses interaction between the query-passage pair, resulting in inferior accuracy. To enhance the performance of dense retrieval models without loss of efficiency, we propose a GNN-encoder model in which query (passage) information is fused into passage (query) representations via graph neural networks that are constructed by queries and their top retrieved passages. By this means, we maintain a dual-encoder structure, and retain some interaction information between query-passage pairs in their representations, which enables us to achieve both efficiency and efficacy in passage retrieval. Evaluation results indicate that our method significantly outperforms the existing models on MSMARCO, Natural Questions and TriviaQA datasets, and achieves the new state-of-the-art on these datasets.