@inproceedings{seo-kim-2026-natural,
title = "Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition",
author = "Seo, Sukmin and
Kim, Geewook",
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.2112/",
pages = "48044--48056",
ISBN = "979-8-89176-459-0",
abstract = "Temporal grounding{---}returning the interval [t{\_}s, t{\_}e] for a natural-language query over a video{---}is the language interface to long-form video, yet has been mainly studied on non-hour-scale videos; the dynamics of hour-scale natural-language grounding remain underexplored. We take the position that at hour-scale, the binding constraint is search, not recognition: Video-LLMs are bottlenecked not by localizing a nearby event, but by searching for the relevant region of a long video. To test this, we release ExtremeWhenBench (2,273 queries over 194 videos, mean 75.7 min, max 9 hr). It combines three properties that no prior benchmark has together: hour-scale mean video length, open-form explicit natural-language queries, and open access that works with standard Video-LLM toolkits. Every open Video-LLM collapses while a frame-level retrieval baseline outperforms them; a failure taxonomy attributes 88{\%} of failures to search; and a retrieve-then-ground hybrid recovers 6.7{\texttimes} over the monolithic Video-LLM{---}mirroring retrieve-then-read in open-domain QA."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="seo-kim-2026-natural">
<titleInfo>
<title>Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition</title>
</titleInfo>
<name type="personal">
<namePart type="given">Sukmin</namePart>
<namePart type="family">Seo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Geewook</namePart>
<namePart type="family">Kim</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>Temporal grounding—returning the interval [t_s, t_e] for a natural-language query over a video—is the language interface to long-form video, yet has been mainly studied on non-hour-scale videos; the dynamics of hour-scale natural-language grounding remain underexplored. We take the position that at hour-scale, the binding constraint is search, not recognition: Video-LLMs are bottlenecked not by localizing a nearby event, but by searching for the relevant region of a long video. To test this, we release ExtremeWhenBench (2,273 queries over 194 videos, mean 75.7 min, max 9 hr). It combines three properties that no prior benchmark has together: hour-scale mean video length, open-form explicit natural-language queries, and open access that works with standard Video-LLM toolkits. Every open Video-LLM collapses while a frame-level retrieval baseline outperforms them; a failure taxonomy attributes 88% of failures to search; and a retrieve-then-ground hybrid recovers 6.7× over the monolithic Video-LLM—mirroring retrieve-then-read in open-domain QA.</abstract>
<identifier type="citekey">seo-kim-2026-natural</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2112/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>48044</start>
<end>48056</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition
%A Seo, Sukmin
%A Kim, Geewook
%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 seo-kim-2026-natural
%X Temporal grounding—returning the interval [t_s, t_e] for a natural-language query over a video—is the language interface to long-form video, yet has been mainly studied on non-hour-scale videos; the dynamics of hour-scale natural-language grounding remain underexplored. We take the position that at hour-scale, the binding constraint is search, not recognition: Video-LLMs are bottlenecked not by localizing a nearby event, but by searching for the relevant region of a long video. To test this, we release ExtremeWhenBench (2,273 queries over 194 videos, mean 75.7 min, max 9 hr). It combines three properties that no prior benchmark has together: hour-scale mean video length, open-form explicit natural-language queries, and open access that works with standard Video-LLM toolkits. Every open Video-LLM collapses while a frame-level retrieval baseline outperforms them; a failure taxonomy attributes 88% of failures to search; and a retrieve-then-ground hybrid recovers 6.7× over the monolithic Video-LLM—mirroring retrieve-then-read in open-domain QA.
%U https://aclanthology.org/2026.emnlp-main.2112/
%P 48044-48056
Markdown (Informal)
[Natural-Language Temporal Grounding in Hour-Long Videos is a Search Problem: A Benchmark and Empirical Decomposition](https://aclanthology.org/2026.emnlp-main.2112/) (Seo & Kim, EMNLP 2026)
ACL