@inproceedings{luo-etal-2026-shift,
title = "{SHIFT}: Self-reconstruction Harnesses Implicit Fine-grained Thinking for Retrieval",
author = "Luo, Yuxiao and
Li, Da and
Zhang, Mingjie and
he, Zhentao and
Zhang, Shikun and
Ye, Wei",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.1182/",
pages = "27103--27115",
ISBN = "979-8-89176-459-0",
abstract = "LLM-based retrievers have become a fundamental component of modern information retrieval systems. The paradigm of ``rewrite-then-retrieve'' introduces explicit reasoning before retrieval. In addition, implicit-reasoning retrievers such as GIRCSE and LaSER improve efficiency by replacing explicit reasoning with soft tokens. Although these methods demonstrated competitive performance on reasoning-intensive retrieval benchmarks, they struggle to address the mismatch between the objectives of retrieval and generation. In this work, we propose SHIFT (**S**elf-reconstruction **H**arnesses **I**mplicit **F**ine-grained **T**hinking for Retrieval), a retrieval training framework based on LLMs. Firstly, we transfer LLMs into reasoning-efficient retrievers with residual projection and task-oriented bidirectional attention aggregation in the latent space. Secondly, we alleviate the mismatch between contrastive learning and implicit reasoning using fine-grained next-token-prediction-based reconstruction. Extensive experiments on reasoning-intensive retrieval benchmarks show that SHIFT consistently outperforms other widely used retrievers. We also carried out a detailed analysis to illustrate how our method works."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="luo-etal-2026-shift">
<titleInfo>
<title>SHIFT: Self-reconstruction Harnesses Implicit Fine-grained Thinking for Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yuxiao</namePart>
<namePart type="family">Luo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Da</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Mingjie</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zhentao</namePart>
<namePart type="family">he</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shikun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Wei</namePart>
<namePart type="family">Ye</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>LLM-based retrievers have become a fundamental component of modern information retrieval systems. The paradigm of “rewrite-then-retrieve” introduces explicit reasoning before retrieval. In addition, implicit-reasoning retrievers such as GIRCSE and LaSER improve efficiency by replacing explicit reasoning with soft tokens. Although these methods demonstrated competitive performance on reasoning-intensive retrieval benchmarks, they struggle to address the mismatch between the objectives of retrieval and generation. In this work, we propose SHIFT (**S**elf-reconstruction **H**arnesses **I**mplicit **F**ine-grained **T**hinking for Retrieval), a retrieval training framework based on LLMs. Firstly, we transfer LLMs into reasoning-efficient retrievers with residual projection and task-oriented bidirectional attention aggregation in the latent space. Secondly, we alleviate the mismatch between contrastive learning and implicit reasoning using fine-grained next-token-prediction-based reconstruction. Extensive experiments on reasoning-intensive retrieval benchmarks show that SHIFT consistently outperforms other widely used retrievers. We also carried out a detailed analysis to illustrate how our method works.</abstract>
<identifier type="citekey">luo-etal-2026-shift</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1182/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>27103</start>
<end>27115</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T SHIFT: Self-reconstruction Harnesses Implicit Fine-grained Thinking for Retrieval
%A Luo, Yuxiao
%A Li, Da
%A Zhang, Mingjie
%A he, Zhentao
%A Zhang, Shikun
%A Ye, Wei
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F luo-etal-2026-shift
%X LLM-based retrievers have become a fundamental component of modern information retrieval systems. The paradigm of “rewrite-then-retrieve” introduces explicit reasoning before retrieval. In addition, implicit-reasoning retrievers such as GIRCSE and LaSER improve efficiency by replacing explicit reasoning with soft tokens. Although these methods demonstrated competitive performance on reasoning-intensive retrieval benchmarks, they struggle to address the mismatch between the objectives of retrieval and generation. In this work, we propose SHIFT (**S**elf-reconstruction **H**arnesses **I**mplicit **F**ine-grained **T**hinking for Retrieval), a retrieval training framework based on LLMs. Firstly, we transfer LLMs into reasoning-efficient retrievers with residual projection and task-oriented bidirectional attention aggregation in the latent space. Secondly, we alleviate the mismatch between contrastive learning and implicit reasoning using fine-grained next-token-prediction-based reconstruction. Extensive experiments on reasoning-intensive retrieval benchmarks show that SHIFT consistently outperforms other widely used retrievers. We also carried out a detailed analysis to illustrate how our method works.
%U https://aclanthology.org/2026.emnlp-main.1182/
%P 27103-27115
Markdown (Informal)
[SHIFT: Self-reconstruction Harnesses Implicit Fine-grained Thinking for Retrieval](https://aclanthology.org/2026.emnlp-main.1182/) (Luo et al., EMNLP 2026)
ACL