Importance-Awareness Masking Network for Robust Document Retrieval
Junping Liu, Jiaqi He, Xinrong Hu, Wangli Yang, Jie Yang, Yi Guo
Abstract
In this paper, we introduce the IMPortance-awaReness maskIng NeTwork (IMPRINT), a novel approach to enhance the robustness of document retrieval systems against query variations, particularly those containing misspellings. Unlike previous models that treat all query components (words/features) equally, IMPRINT prioritizes the most important components while masking out less relevant ones. Specifically, we propose a Mutual Information-based measure to quantify component importance and integrate it into a dynamic masking mechanism that adjusts the retention probability of each component. Our method is evaluated on a combination of three benchmark datasets and three types of query variations. The experimental results show substantial performance gains compared to state-of-the-art models, achieving an average improvement of 1.2 absolute MRR@10 points in retrieval accuracy.
BibTeX
@inproceedings{icassp2025_importanceawaren,
title = {Importance-Awareness Masking Network for Robust Document Retrieval},
author = {Junping Liu and Jiaqi He and Xinrong Hu and Wangli Yang and Jie Yang and Yi Guo},
booktitle = {ICASSP 2025},
year = {2025}
}