arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2302.06589cs.IRcs.CL
收藏

利用上下文化后期交互改进神经网络重排序器的分布外泛化

Improving Out-of-Distribution Generalization of Neural Rerankers with Contextualized Late Interaction

University of Waterloo(滑铁卢大学)

Xinyu Zhang, Minghan Li, Jimmy Lin

摘要:

信息检索的最新进展发现,将查询和文档表示嵌入到多向量中可以在分布外的数据集上产生鲁棒的双编码器检索器。在本文中,我们探讨了后期交互(最简单的多向量形式)是否也有助于仅使用[CLS]向量来计算相似性得分的神经重排序器。虽然直观地说,前几层的重排序器的注意力机制已经收集了令牌级别的信息,但我们发现添加后期交互仍然会在分布外数据集上带来额外的5%的平均改善,延迟几乎没有增加,域内有效性也没有下降。通过大量的实验和分析,我们发现,在不同的模型大小和不同性质的第一阶段检索器是一致的,并且在较长的查询中,这种改进更为突出。

英文摘要:

Recent progress in information retrieval finds that embedding query and document representation into multi-vector yields a robust bi-encoder retriever on out-of-distribution datasets. In this paper, we explore whether late interaction, the simplest form of multi-vector, is also helpful to neural rerankers that only use the [CLS] vector to compute the similarity score. Although intuitively, the attention mechanism of rerankers at the previous layers already gathers the token-level information, we find adding late interaction still brings an extra 5% improvement in average on out-of-distribution datasets, with little increase in latency and no degradation in in-domain effectiveness. Through extensive experiments and analysis, we show that the finding is consistent across different model sizes and first-stage retrievers of diverse natures and that the improvement is more prominent on longer queries.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑