@inproceedings{tao-etal-2026-personalizing,
title = "Personalizing How Recommendation Agents Reason: Self-Evolving Per-User Skills for Recommendation Agents",
author = "Tao, Zhen and
Lai, Riwei and
Yu, Chenyun and
Chen, Weixin and
Chen, Li and
Kong, Beibei and
Cheng, Lei and
Zhuo, Chengxiang and
Li, Zang and
Sun, Qingqiang",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {EMNLP} 2026",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-emnlp.872/",
pages = "18438--18460",
ISBN = "979-8-89176-462-0",
abstract = "LLM-based recommendation agents personalize what they know through evolving per-user semantic memory, yet how they perform reasoning relies on a universal, static system prompt shared across all users. When a recommendation fails, the agent only updates its memory, but never examines or revises the reasoning that leads to the failure. To close this gap, we propose SAGER (Self-Evolving Agent for Personalized Recommendation), the first framework to equip each user with a dedicated, continuously evolving policy skill. SAGER incorporates three tailored mechanisms for recommendation scenarios: (i) a two-representation skill architecture constrained by a Cognitive Injection Boundary, decoupling a rich evolution substrate from a minimal inference-time injection; (ii) a contrastive differential chain-of-thought engine that diagnoses ranking errors against unchosen near-misses under a refine-not-replace principle; and (iii) skill-augmented listwise reasoning that mitigates the Score-Tie Bottleneck, enabling the slim skill to serve as an active tie-breaker. Evaluated on four public benchmarks, SAGER achieves state-of-the-art accuracy. Furthermore, plugging its policy skill layer into three structurally distinct host agents delivers consistent 9{\%}{--}12{\%} gains on the H@1 metric. These results indicate that policy skill evolution provides architecture-agnostic improvements that are largely complementary to memory accumulation."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="tao-etal-2026-personalizing">
<titleInfo>
<title>Personalizing How Recommendation Agents Reason: Self-Evolving Per-User Skills for Recommendation Agents</title>
</titleInfo>
<name type="personal">
<namePart type="given">Zhen</namePart>
<namePart type="family">Tao</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Riwei</namePart>
<namePart type="family">Lai</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chenyun</namePart>
<namePart type="family">Yu</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Weixin</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Li</namePart>
<namePart type="family">Chen</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Beibei</namePart>
<namePart type="family">Kong</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Lei</namePart>
<namePart type="family">Cheng</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Chengxiang</namePart>
<namePart type="family">Zhuo</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Zang</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Qingqiang</namePart>
<namePart type="family">Sun</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>Findings of the Association for Computational Linguistics: EMNLP 2026</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">979-8-89176-462-0</identifier>
</relatedItem>
<abstract>LLM-based recommendation agents personalize what they know through evolving per-user semantic memory, yet how they perform reasoning relies on a universal, static system prompt shared across all users. When a recommendation fails, the agent only updates its memory, but never examines or revises the reasoning that leads to the failure. To close this gap, we propose SAGER (Self-Evolving Agent for Personalized Recommendation), the first framework to equip each user with a dedicated, continuously evolving policy skill. SAGER incorporates three tailored mechanisms for recommendation scenarios: (i) a two-representation skill architecture constrained by a Cognitive Injection Boundary, decoupling a rich evolution substrate from a minimal inference-time injection; (ii) a contrastive differential chain-of-thought engine that diagnoses ranking errors against unchosen near-misses under a refine-not-replace principle; and (iii) skill-augmented listwise reasoning that mitigates the Score-Tie Bottleneck, enabling the slim skill to serve as an active tie-breaker. Evaluated on four public benchmarks, SAGER achieves state-of-the-art accuracy. Furthermore, plugging its policy skill layer into three structurally distinct host agents delivers consistent 9%–12% gains on the H@1 metric. These results indicate that policy skill evolution provides architecture-agnostic improvements that are largely complementary to memory accumulation.</abstract>
<identifier type="citekey">tao-etal-2026-personalizing</identifier>
<location>
<url>https://aclanthology.org/2026.findings-emnlp.872/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>18438</start>
<end>18460</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Personalizing How Recommendation Agents Reason: Self-Evolving Per-User Skills for Recommendation Agents
%A Tao, Zhen
%A Lai, Riwei
%A Yu, Chenyun
%A Chen, Weixin
%A Chen, Li
%A Kong, Beibei
%A Cheng, Lei
%A Zhuo, Chengxiang
%A Li, Zang
%A Sun, Qingqiang
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Findings of the Association for Computational Linguistics: EMNLP 2026
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-462-0
%F tao-etal-2026-personalizing
%X LLM-based recommendation agents personalize what they know through evolving per-user semantic memory, yet how they perform reasoning relies on a universal, static system prompt shared across all users. When a recommendation fails, the agent only updates its memory, but never examines or revises the reasoning that leads to the failure. To close this gap, we propose SAGER (Self-Evolving Agent for Personalized Recommendation), the first framework to equip each user with a dedicated, continuously evolving policy skill. SAGER incorporates three tailored mechanisms for recommendation scenarios: (i) a two-representation skill architecture constrained by a Cognitive Injection Boundary, decoupling a rich evolution substrate from a minimal inference-time injection; (ii) a contrastive differential chain-of-thought engine that diagnoses ranking errors against unchosen near-misses under a refine-not-replace principle; and (iii) skill-augmented listwise reasoning that mitigates the Score-Tie Bottleneck, enabling the slim skill to serve as an active tie-breaker. Evaluated on four public benchmarks, SAGER achieves state-of-the-art accuracy. Furthermore, plugging its policy skill layer into three structurally distinct host agents delivers consistent 9%–12% gains on the H@1 metric. These results indicate that policy skill evolution provides architecture-agnostic improvements that are largely complementary to memory accumulation.
%U https://aclanthology.org/2026.findings-emnlp.872/
%P 18438-18460
Markdown (Informal)
[Personalizing How Recommendation Agents Reason: Self-Evolving Per-User Skills for Recommendation Agents](https://aclanthology.org/2026.findings-emnlp.872/) (Tao et al., Findings 2026)
ACL
- Zhen Tao, Riwei Lai, Chenyun Yu, Weixin Chen, Li Chen, Beibei Kong, Lei Cheng, Chengxiang Zhuo, Zang Li, and Qingqiang Sun. 2026. Personalizing How Recommendation Agents Reason: Self-Evolving Per-User Skills for Recommendation Agents. In Findings of the Association for Computational Linguistics: EMNLP 2026, pages 18438–18460, Budapest, Hungary. Association for Computational Linguistics.