@inproceedings{gangopadhyay-etal-2026-reflect,
title = "Reflect and Retrieve: Multimodal Agentic Retrieval with Reflection and Query Agents for Document Search",
author = "Gangopadhyay, Tryambak and
Salekin, Md Sirajus and
Qian, Yiyue and
Agu, Nkechinyere Nneka and
Zhang, Shinan and
Han, Rinne",
editor = "T.Y.S.S., Santosh and
Agarwal, Amit and
Gupta, Vivek and
Kementchedjhieva, Yova and
Garimella, Aparna and
Hou, Yufang",
booktitle = "The Proceedings of Workshop on Document Intelligence and Understanding",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.docinsights-1.10/",
pages = "138--154",
ISBN = "979-8-89176-477-4",
abstract = "Retrieving relevant pages from visually rich documents, with evidence in charts, tables, and figures rather than body text, remains difficult: a single-pass query against a fixed index misses content in visual elements, and reordering the same candidates cannot recover a page never retrieved. We propose Reflect and Retrieve, an agentic retrieval approach centered on a Reflection Agent. After an initial page-level retrieval, the agent uses a multimodal large language model to visually reason over the retrieved page images, judge whether the evidence is sufficient, and, when gaps remain, an orchestrator issues targeted follow-up retrievals to recover the missing pages. Evaluating on two benchmarks, MMDocIR and ViDoRe V3, the Reflection Agent delivers the dominant agentic gain: it is the only configuration to beat single-step retrieval for every reasoning model on both, improving Recall@N from 0.628 to 0.663 on MMDocIR and 0.566 to 0.644 on ViDoRe V3. The improvement comes from recovering relevant pages absent in initial retrieval. Our approach indicates a clear design order: selecting an optimal embedding approach for the retrieval tool is the first step, after which the Reflection Agent contributes more value than a query decomposition agent, positioning visual reflection as the key component for document search."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="gangopadhyay-etal-2026-reflect">
<titleInfo>
<title>Reflect and Retrieve: Multimodal Agentic Retrieval with Reflection and Query Agents for Document Search</title>
</titleInfo>
<name type="personal">
<namePart type="given">Tryambak</namePart>
<namePart type="family">Gangopadhyay</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Md</namePart>
<namePart type="given">Sirajus</namePart>
<namePart type="family">Salekin</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yiyue</namePart>
<namePart type="family">Qian</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Nkechinyere</namePart>
<namePart type="given">Nneka</namePart>
<namePart type="family">Agu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Shinan</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Rinne</namePart>
<namePart type="family">Han</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>The Proceedings of Workshop on Document Intelligence and Understanding</title>
</titleInfo>
<name type="personal">
<namePart type="given">Santosh</namePart>
<namePart type="family">T.Y.S.S.</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Amit</namePart>
<namePart type="family">Agarwal</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Vivek</namePart>
<namePart type="family">Gupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yova</namePart>
<namePart type="family">Kementchedjhieva</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Aparna</namePart>
<namePart type="family">Garimella</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yufang</namePart>
<namePart type="family">Hou</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">979-8-89176-477-4</identifier>
</relatedItem>
<abstract>Retrieving relevant pages from visually rich documents, with evidence in charts, tables, and figures rather than body text, remains difficult: a single-pass query against a fixed index misses content in visual elements, and reordering the same candidates cannot recover a page never retrieved. We propose Reflect and Retrieve, an agentic retrieval approach centered on a Reflection Agent. After an initial page-level retrieval, the agent uses a multimodal large language model to visually reason over the retrieved page images, judge whether the evidence is sufficient, and, when gaps remain, an orchestrator issues targeted follow-up retrievals to recover the missing pages. Evaluating on two benchmarks, MMDocIR and ViDoRe V3, the Reflection Agent delivers the dominant agentic gain: it is the only configuration to beat single-step retrieval for every reasoning model on both, improving Recall@N from 0.628 to 0.663 on MMDocIR and 0.566 to 0.644 on ViDoRe V3. The improvement comes from recovering relevant pages absent in initial retrieval. Our approach indicates a clear design order: selecting an optimal embedding approach for the retrieval tool is the first step, after which the Reflection Agent contributes more value than a query decomposition agent, positioning visual reflection as the key component for document search.</abstract>
<identifier type="citekey">gangopadhyay-etal-2026-reflect</identifier>
<location>
<url>https://aclanthology.org/2026.docinsights-1.10/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>138</start>
<end>154</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Reflect and Retrieve: Multimodal Agentic Retrieval with Reflection and Query Agents for Document Search
%A Gangopadhyay, Tryambak
%A Salekin, Md Sirajus
%A Qian, Yiyue
%A Agu, Nkechinyere Nneka
%A Zhang, Shinan
%A Han, Rinne
%Y T.Y.S.S., Santosh
%Y Agarwal, Amit
%Y Gupta, Vivek
%Y Kementchedjhieva, Yova
%Y Garimella, Aparna
%Y Hou, Yufang
%S The Proceedings of Workshop on Document Intelligence and Understanding
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-477-4
%F gangopadhyay-etal-2026-reflect
%X Retrieving relevant pages from visually rich documents, with evidence in charts, tables, and figures rather than body text, remains difficult: a single-pass query against a fixed index misses content in visual elements, and reordering the same candidates cannot recover a page never retrieved. We propose Reflect and Retrieve, an agentic retrieval approach centered on a Reflection Agent. After an initial page-level retrieval, the agent uses a multimodal large language model to visually reason over the retrieved page images, judge whether the evidence is sufficient, and, when gaps remain, an orchestrator issues targeted follow-up retrievals to recover the missing pages. Evaluating on two benchmarks, MMDocIR and ViDoRe V3, the Reflection Agent delivers the dominant agentic gain: it is the only configuration to beat single-step retrieval for every reasoning model on both, improving Recall@N from 0.628 to 0.663 on MMDocIR and 0.566 to 0.644 on ViDoRe V3. The improvement comes from recovering relevant pages absent in initial retrieval. Our approach indicates a clear design order: selecting an optimal embedding approach for the retrieval tool is the first step, after which the Reflection Agent contributes more value than a query decomposition agent, positioning visual reflection as the key component for document search.
%U https://aclanthology.org/2026.docinsights-1.10/
%P 138-154
Markdown (Informal)
[Reflect and Retrieve: Multimodal Agentic Retrieval with Reflection and Query Agents for Document Search](https://aclanthology.org/2026.docinsights-1.10/) (Gangopadhyay et al., DocInsights 2026)
ACL