Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Yang, Zhen, Shao, Zhou, Dong, Yuxiao, Tang, Jie
Format: Preprint
Veröffentlicht: 2024
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2402.11855
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866916130986131456
author Yang, Zhen
Shao, Zhou
Dong, Yuxiao
Tang, Jie
author_facet Yang, Zhen
Shao, Zhou
Dong, Yuxiao
Tang, Jie
contents Negative sampling stands as a pivotal technique in dense retrieval, essential for training effective retrieval models and significantly impacting retrieval performance. While existing negative sampling methods have made commendable progress by leveraging hard negatives, a comprehensive guiding principle for constructing negative candidates and designing negative sampling distributions is still lacking. To bridge this gap, we embark on a theoretical analysis of negative sampling in dense retrieval. This exploration culminates in the unveiling of the quasi-triangular principle, a novel framework that elucidates the triangular-like interplay between query, positive document, and negative document. Fueled by this guiding principle, we introduce TriSampler, a straightforward yet highly effective negative sampling method. The keypoint of TriSampler lies in its ability to selectively sample more informative negatives within a prescribed constrained region. Experimental evaluation show that TriSampler consistently attains superior retrieval performance across a diverse of representative retrieval models.
format Preprint
id arxiv_https___arxiv_org_abs_2402_11855
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle TriSampler: A Better Negative Sampling Principle for Dense Retrieval
Yang, Zhen
Shao, Zhou
Dong, Yuxiao
Tang, Jie
Information Retrieval
Negative sampling stands as a pivotal technique in dense retrieval, essential for training effective retrieval models and significantly impacting retrieval performance. While existing negative sampling methods have made commendable progress by leveraging hard negatives, a comprehensive guiding principle for constructing negative candidates and designing negative sampling distributions is still lacking. To bridge this gap, we embark on a theoretical analysis of negative sampling in dense retrieval. This exploration culminates in the unveiling of the quasi-triangular principle, a novel framework that elucidates the triangular-like interplay between query, positive document, and negative document. Fueled by this guiding principle, we introduce TriSampler, a straightforward yet highly effective negative sampling method. The keypoint of TriSampler lies in its ability to selectively sample more informative negatives within a prescribed constrained region. Experimental evaluation show that TriSampler consistently attains superior retrieval performance across a diverse of representative retrieval models.
title TriSampler: A Better Negative Sampling Principle for Dense Retrieval
topic Information Retrieval
url https://arxiv.org/abs/2402.11855