@inproceedings{ahn-etal-2026-trace,
title = "{TRACE}: A Long-Term Conversational Recommendation Dataset with Evolving User Personas",
author = "Ahn, Jihyung and
Han, Seonah and
Park, Chanho and
IM, Hyeongwoo and
Kim, Daeyeong and
Kim, Subin and
Chang, Buru",
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.1804/",
pages = "41088--41136",
ISBN = "979-8-89176-459-0",
abstract = "Conversational Recommender Systems (CRS) have improved personalized recommendation through dialogue, but most studies and datasets remain limited to single-session settings. Such single-session settings make it difficult to capture long-term preference evolution and the interplay between latent behavioral patterns and explicit conversational styles, partly because real-world conversations aligned with users' long-term behavioral histories are scarce. To address this gap, we introduce TRACE, a long-term CRS dataset built from behavioral histories that reflect persistent user preferences. Our analysis shows that TRACE provides reliable and diverse behavioral personas, as well as higher-quality CRS interactions than existing CRS datasets. Building on TRACE, we introduce EvoCRS, a persona-based CRS framework that models accumulated and evolving preferences through updated behavioral and dialogue personas. Experiments show that both persona types improve recommendation performance, and that persona updates are important for capturing long-term preference evolution. Our code and datasets are available online."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="ahn-etal-2026-trace">
<titleInfo>
<title>TRACE: A Long-Term Conversational Recommendation Dataset with Evolving User Personas</title>
</titleInfo>
<name type="personal">
<namePart type="given">Jihyung</namePart>
<namePart type="family">Ahn</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Seonah</namePart>
<namePart type="family">Han</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chanho</namePart>
<namePart type="family">Park</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Hyeongwoo</namePart>
<namePart type="family">IM</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Daeyeong</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Subin</namePart>
<namePart type="family">Kim</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Buru</namePart>
<namePart type="family">Chang</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>Conversational Recommender Systems (CRS) have improved personalized recommendation through dialogue, but most studies and datasets remain limited to single-session settings. Such single-session settings make it difficult to capture long-term preference evolution and the interplay between latent behavioral patterns and explicit conversational styles, partly because real-world conversations aligned with users’ long-term behavioral histories are scarce. To address this gap, we introduce TRACE, a long-term CRS dataset built from behavioral histories that reflect persistent user preferences. Our analysis shows that TRACE provides reliable and diverse behavioral personas, as well as higher-quality CRS interactions than existing CRS datasets. Building on TRACE, we introduce EvoCRS, a persona-based CRS framework that models accumulated and evolving preferences through updated behavioral and dialogue personas. Experiments show that both persona types improve recommendation performance, and that persona updates are important for capturing long-term preference evolution. Our code and datasets are available online.</abstract>
<identifier type="citekey">ahn-etal-2026-trace</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.1804/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>41088</start>
<end>41136</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T TRACE: A Long-Term Conversational Recommendation Dataset with Evolving User Personas
%A Ahn, Jihyung
%A Han, Seonah
%A Park, Chanho
%A IM, Hyeongwoo
%A Kim, Daeyeong
%A Kim, Subin
%A Chang, Buru
%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 ahn-etal-2026-trace
%X Conversational Recommender Systems (CRS) have improved personalized recommendation through dialogue, but most studies and datasets remain limited to single-session settings. Such single-session settings make it difficult to capture long-term preference evolution and the interplay between latent behavioral patterns and explicit conversational styles, partly because real-world conversations aligned with users’ long-term behavioral histories are scarce. To address this gap, we introduce TRACE, a long-term CRS dataset built from behavioral histories that reflect persistent user preferences. Our analysis shows that TRACE provides reliable and diverse behavioral personas, as well as higher-quality CRS interactions than existing CRS datasets. Building on TRACE, we introduce EvoCRS, a persona-based CRS framework that models accumulated and evolving preferences through updated behavioral and dialogue personas. Experiments show that both persona types improve recommendation performance, and that persona updates are important for capturing long-term preference evolution. Our code and datasets are available online.
%U https://aclanthology.org/2026.emnlp-main.1804/
%P 41088-41136
Markdown (Informal)
[TRACE: A Long-Term Conversational Recommendation Dataset with Evolving User Personas](https://aclanthology.org/2026.emnlp-main.1804/) (Ahn et al., EMNLP 2026)
ACL