@inproceedings{pan-etal-2026-empirical,
title = "An Empirical Study of Position Bias in Visualized Text Retrieval",
author = "Pan, Cuiqiaoshu and
Zeng, Ziyang and
Zhang, Dun and
Yang, Yuqing",
editor = "Zhang, Yue and
Xu, Ruifeng and
Purwarianti, Ayu and
Rong, Tong and
Suzuki, Jun",
booktitle = "Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the {A}sia-Pacific Chapter of the {A}ssociation for {C}omputational {L}inguistics (Volume 1: Long Papers)",
month = nov,
year = "2026",
address = "Hengqin, China",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.aacl-long.259/",
pages = "5740--5760",
ISBN = "979-8-89176-439-2",
abstract = "Visualized text retrieval, which renders text into images and retrieves them with vision-language models, has been proposed to avoid lossy text parsing and compress long contexts. We ask whether it also bypasses a known flaw of text retrieval: position bias, the tendency to overweight a passage{'}s beginning. Rendering 1D text into a 2D image may change the bias in two respects: its shape, which position is favored, and its magnitude, how strongly performance varies across positions. We test this by rendering the PosIR benchmark as controlled plain text images and comparing eight visual retrievers with four text retrievers on identical instances. We find that (1) position bias persists, with all visual retrievers exhibiting measurable bias (macro PSI 0.061{--}0.168); (2) its shape is transformed: seven of eight visual retrievers favor the middle of the page in an inverted-U pattern, contrasting with both the primacy bias of text retrievers and the lost-in-the-middle effect of LLMs with long contexts; (3) its magnitude is sensitive to rendering settings, as taller images amplify PSI and can even reverse positional preference, while font size has a similarly large effect; and (4) under matched backbones, visual retrievers exhibit lower position bias than their text counterparts on longer documents, but this advantage vanishes beyond the pixel budget and at document scale. Visualized text retrieval therefore does not eliminate position bias; it transforms a familiar 1D textual bias into a more complex 2D spatial bias."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="pan-etal-2026-empirical">
<titleInfo>
<title>An Empirical Study of Position Bias in Visualized Text Retrieval</title>
</titleInfo>
<name type="personal">
<namePart type="given">Cuiqiaoshu</namePart>
<namePart type="family">Pan</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ziyang</namePart>
<namePart type="family">Zeng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Dun</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Yuqing</namePart>
<namePart type="family">Yang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-11</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)</title>
</titleInfo>
<name type="personal">
<namePart type="given">Yue</namePart>
<namePart type="family">Zhang</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ruifeng</namePart>
<namePart type="family">Xu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Ayu</namePart>
<namePart type="family">Purwarianti</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tong</namePart>
<namePart type="family">Rong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jun</namePart>
<namePart type="family">Suzuki</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Hengqin, China</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-439-2</identifier>
</relatedItem>
<abstract>Visualized text retrieval, which renders text into images and retrieves them with vision-language models, has been proposed to avoid lossy text parsing and compress long contexts. We ask whether it also bypasses a known flaw of text retrieval: position bias, the tendency to overweight a passage’s beginning. Rendering 1D text into a 2D image may change the bias in two respects: its shape, which position is favored, and its magnitude, how strongly performance varies across positions. We test this by rendering the PosIR benchmark as controlled plain text images and comparing eight visual retrievers with four text retrievers on identical instances. We find that (1) position bias persists, with all visual retrievers exhibiting measurable bias (macro PSI 0.061–0.168); (2) its shape is transformed: seven of eight visual retrievers favor the middle of the page in an inverted-U pattern, contrasting with both the primacy bias of text retrievers and the lost-in-the-middle effect of LLMs with long contexts; (3) its magnitude is sensitive to rendering settings, as taller images amplify PSI and can even reverse positional preference, while font size has a similarly large effect; and (4) under matched backbones, visual retrievers exhibit lower position bias than their text counterparts on longer documents, but this advantage vanishes beyond the pixel budget and at document scale. Visualized text retrieval therefore does not eliminate position bias; it transforms a familiar 1D textual bias into a more complex 2D spatial bias.</abstract>
<identifier type="citekey">pan-etal-2026-empirical</identifier>
<location>
<url>https://aclanthology.org/2026.aacl-long.259/</url>
</location>
<part>
<date>2026-11</date>
<extent unit="page">
<start>5740</start>
<end>5760</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T An Empirical Study of Position Bias in Visualized Text Retrieval
%A Pan, Cuiqiaoshu
%A Zeng, Ziyang
%A Zhang, Dun
%A Yang, Yuqing
%Y Zhang, Yue
%Y Xu, Ruifeng
%Y Purwarianti, Ayu
%Y Rong, Tong
%Y Suzuki, Jun
%S Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2026
%8 November
%I Association for Computational Linguistics
%C Hengqin, China
%@ 979-8-89176-439-2
%F pan-etal-2026-empirical
%X Visualized text retrieval, which renders text into images and retrieves them with vision-language models, has been proposed to avoid lossy text parsing and compress long contexts. We ask whether it also bypasses a known flaw of text retrieval: position bias, the tendency to overweight a passage’s beginning. Rendering 1D text into a 2D image may change the bias in two respects: its shape, which position is favored, and its magnitude, how strongly performance varies across positions. We test this by rendering the PosIR benchmark as controlled plain text images and comparing eight visual retrievers with four text retrievers on identical instances. We find that (1) position bias persists, with all visual retrievers exhibiting measurable bias (macro PSI 0.061–0.168); (2) its shape is transformed: seven of eight visual retrievers favor the middle of the page in an inverted-U pattern, contrasting with both the primacy bias of text retrievers and the lost-in-the-middle effect of LLMs with long contexts; (3) its magnitude is sensitive to rendering settings, as taller images amplify PSI and can even reverse positional preference, while font size has a similarly large effect; and (4) under matched backbones, visual retrievers exhibit lower position bias than their text counterparts on longer documents, but this advantage vanishes beyond the pixel budget and at document scale. Visualized text retrieval therefore does not eliminate position bias; it transforms a familiar 1D textual bias into a more complex 2D spatial bias.
%U https://aclanthology.org/2026.aacl-long.259/
%P 5740-5760
Markdown (Informal)
[An Empirical Study of Position Bias in Visualized Text Retrieval](https://aclanthology.org/2026.aacl-long.259/) (Pan et al., AACL-IJCNLP 2026)
ACL
- Cuiqiaoshu Pan, Ziyang Zeng, Dun Zhang, and Yuqing Yang. 2026. An Empirical Study of Position Bias in Visualized Text Retrieval. In Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (Volume 1: Long Papers), pages 5740–5760, Hengqin, China. Association for Computational Linguistics.