Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent Article Swipe

PDF

Xueyang Feng , Jingsen Zhang , Jiakai Tang , Wei Li , Gengyuan Cai , Xu Chen , Quanyu Dai , Yuemin Zhu , Zhenhua Dong ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2506.14302

Recent advancements in Large Language Models (LLMs) have significantly propelled the development of Conversational Recommendation Agents (CRAs). However, these agents often generate short-sighted responses that fail to sustain user guidance and meet expectations. Although preference optimization has proven effective in aligning LLMs with user expectations, it remains costly and performs poorly in multi-turn dialogue. To address this challenge, we introduce a novel multi-turn preference optimization (MTPO) paradigm ECPO, which leverages Expectation Confirmation Theory to explicitly model the evolution of user satisfaction throughout multi-turn dialogues, uncovering the underlying causes of dissatisfaction. These causes can be utilized to support targeted optimization of unsatisfactory responses, thereby achieving turn-level preference optimization. ECPO ingeniously eliminates the significant sampling overhead of existing MTPO methods while ensuring the optimization process drives meaningful improvements. To support ECPO, we introduce an LLM-based user simulator, AILO, to simulate user feedback and perform expectation confirmation during conversational recommendations. Experimental results show that ECPO significantly enhances CRA's interaction capabilities, delivering notable improvements in both efficiency and effectiveness over existing MTPO methods.

Related Topics

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2506.14302
PDF: https://arxiv.org/pdf/2506.14302
OA Status: green
OpenAlex ID: https://openalex.org/W4415311757

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4415311757

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2506.14302

Digital Object Identifier
Title: Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-06-17

Full publication date if available
Authors: Xueyang Feng, Jingsen Zhang, Jiakai Tang, Wei Li, Gengyuan Cai, Xu Chen, Quanyu Dai, Yuemin Zhu, Zhenhua Dong

List of authors in order
Landing page: https://arxiv.org/abs/2506.14302

Publisher landing page
PDF URL: https://arxiv.org/pdf/2506.14302

Direct link to full text PDF
Open access: Yes

Whether a free full text is available
OA status: green

Open access status per OpenAlex
OA URL: https://arxiv.org/pdf/2506.14302

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4415311757
doi	https://doi.org/10.48550/arxiv.2506.14302
ids.doi	https://doi.org/10.48550/arxiv.2506.14302
ids.openalex	https://openalex.org/W4415311757
fwci
type	preprint
title	Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10203
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.8453999757766724
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1710
topics[0].subfield.display_name	Information Systems
topics[0].display_name	Recommender Systems and Techniques
topics[1].id	https://openalex.org/T13083
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.8188999891281128
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1702
topics[1].subfield.display_name	Artificial Intelligence
topics[1].display_name	Advanced Text Analysis Techniques
topics[2].id	https://openalex.org/T10664
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.7638999819755554
topics[2].domain.id	https://openalex.org/domains/3
topics[2].domain.display_name	Physical Sciences
topics[2].subfield.id	https://openalex.org/subfields/1702
topics[2].subfield.display_name	Artificial Intelligence
topics[2].display_name	Sentiment Analysis and Opinion Mining
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2506.14302
locations[0].is_oa	True
locations[0].source.id	https://openalex.org/S4306400194
locations[0].source.issn
locations[0].source.type	repository
locations[0].source.is_oa	True
locations[0].source.issn_l
locations[0].source.is_core	False
locations[0].source.is_in_doaj	False
locations[0].source.display_name	arXiv (Cornell University)
locations[0].source.host_organization	https://openalex.org/I205783295
locations[0].source.host_organization_name	Cornell University
locations[0].source.host_organization_lineage	https://openalex.org/I205783295
locations[0].license
locations[0].pdf_url	https://arxiv.org/pdf/2506.14302
locations[0].version	submittedVersion
locations[0].raw_type	text
locations[0].license_id
locations[0].is_accepted	False
locations[0].is_published	False
locations[0].raw_source_name
locations[0].landing_page_url	http://arxiv.org/abs/2506.14302
locations[1].id	doi:10.48550/arxiv.2506.14302
locations[1].is_oa	True
locations[1].source.id	https://openalex.org/S4306400194
locations[1].source.issn
locations[1].source.type	repository
locations[1].source.is_oa	True
locations[1].source.issn_l
locations[1].source.is_core	False
locations[1].source.is_in_doaj	False
locations[1].source.display_name	arXiv (Cornell University)
locations[1].source.host_organization	https://openalex.org/I205783295
locations[1].source.host_organization_name	Cornell University
locations[1].source.host_organization_lineage	https://openalex.org/I205783295
locations[1].license
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id
locations[1].is_accepted	False
locations[1].is_published
locations[1].raw_source_name
locations[1].landing_page_url	https://doi.org/10.48550/arxiv.2506.14302
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5101929118
authorships[0].author.orcid	https://orcid.org/0000-0003-4426-5732
authorships[0].author.display_name	Xueyang Feng
authorships[0].author_position	first
authorships[0].raw_author_name	Feng, Xueyang
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5101658564
authorships[1].author.orcid	https://orcid.org/0000-0003-2997-3386
authorships[1].author.display_name	Jingsen Zhang
authorships[1].author_position	middle
authorships[1].raw_author_name	Zhang, Jingsen
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5043446092
authorships[2].author.orcid	https://orcid.org/0000-0001-9543-8889
authorships[2].author.display_name	Jiakai Tang
authorships[2].author_position	middle
authorships[2].raw_author_name	Tang, Jiakai
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5100663664
authorships[3].author.orcid	https://orcid.org/0000-0001-7530-3485
authorships[3].author.display_name	Wei Li
authorships[3].author_position	middle
authorships[3].raw_author_name	Li, Wei
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5101849887
authorships[4].author.orcid	https://orcid.org/0000-0002-2854-4478
authorships[4].author.display_name	Gengyuan Cai
authorships[4].author_position	middle
authorships[4].raw_author_name	Cai, Guohao
authorships[4].is_corresponding	False
authorships[5].author.id	https://openalex.org/A5101755392
authorships[5].author.orcid	https://orcid.org/0000-0003-0144-1775
authorships[5].author.display_name	Xu Chen
authorships[5].author_position	middle
authorships[5].raw_author_name	Chen, Xu
authorships[5].is_corresponding	False
authorships[6].author.id	https://openalex.org/A5048783161
authorships[6].author.orcid	https://orcid.org/0000-0001-7578-2738
authorships[6].author.display_name	Quanyu Dai
authorships[6].author_position	middle
authorships[6].raw_author_name	Dai, Quanyu
authorships[6].is_corresponding	False
authorships[7].author.id	https://openalex.org/A5100605235
authorships[7].author.orcid	https://orcid.org/0000-0001-6814-1449
authorships[7].author.display_name	Yuemin Zhu
authorships[7].author_position	middle
authorships[7].raw_author_name	Zhu, Yue
authorships[7].is_corresponding	False
authorships[8].author.id	https://openalex.org/A5028135192
authorships[8].author.orcid	https://orcid.org/0000-0002-5503-4414
authorships[8].author.display_name	Zhenhua Dong
authorships[8].author_position	last
authorships[8].raw_author_name	Dong, Zhenhua
authorships[8].is_corresponding	False
has_content.pdf	True
has_content.grobid_xml	True
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2506.14302
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-18T00:00:00
display_name	Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10203
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.8453999757766724
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1710
primary_topic.subfield.display_name	Information Systems
primary_topic.display_name	Recommender Systems and Techniques
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2506.14302
best_oa_location.is_oa	True
best_oa_location.source.id	https://openalex.org/S4306400194
best_oa_location.source.issn
best_oa_location.source.type	repository
best_oa_location.source.is_oa	True
best_oa_location.source.issn_l
best_oa_location.source.is_core	False
best_oa_location.source.is_in_doaj	False
best_oa_location.source.display_name	arXiv (Cornell University)
best_oa_location.source.host_organization	https://openalex.org/I205783295
best_oa_location.source.host_organization_name	Cornell University
best_oa_location.source.host_organization_lineage	https://openalex.org/I205783295
best_oa_location.license
best_oa_location.pdf_url	https://arxiv.org/pdf/2506.14302
best_oa_location.version	submittedVersion
best_oa_location.raw_type	text
best_oa_location.license_id
best_oa_location.is_accepted	False
best_oa_location.is_published	False
best_oa_location.raw_source_name
best_oa_location.landing_page_url	http://arxiv.org/abs/2506.14302
primary_location.id	pmh:oai:arXiv.org:2506.14302
primary_location.is_oa	True
primary_location.source.id	https://openalex.org/S4306400194
primary_location.source.issn
primary_location.source.type	repository
primary_location.source.is_oa	True
primary_location.source.issn_l
primary_location.source.is_core	False
primary_location.source.is_in_doaj	False
primary_location.source.display_name	arXiv (Cornell University)
primary_location.source.host_organization	https://openalex.org/I205783295
primary_location.source.host_organization_name	Cornell University
primary_location.source.host_organization_lineage	https://openalex.org/I205783295
primary_location.license
primary_location.pdf_url	https://arxiv.org/pdf/2506.14302
primary_location.version	submittedVersion
primary_location.raw_type	text
primary_location.license_id
primary_location.is_accepted	False
primary_location.is_published	False
primary_location.raw_source_name
primary_location.landing_page_url	http://arxiv.org/abs/2506.14302
publication_date	2025-06-17
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	60
abstract_inverted_index.To	54, 126
abstract_inverted_index.an	131
abstract_inverted_index.be	93
abstract_inverted_index.in	2, 39, 51, 160
abstract_inverted_index.it	45
abstract_inverted_index.of	12, 78, 88, 99, 114
abstract_inverted_index.to	26, 73, 95, 136
abstract_inverted_index.we	58, 129
abstract_inverted_index.and	30, 48, 140, 163
abstract_inverted_index.can	92
abstract_inverted_index.has	36
abstract_inverted_index.the	10, 76, 85, 110, 120
abstract_inverted_index.ECPO	107, 151
abstract_inverted_index.LLMs	41
abstract_inverted_index.MTPO	116, 167
abstract_inverted_index.both	161
abstract_inverted_index.fail	25
abstract_inverted_index.have	7
abstract_inverted_index.meet	31
abstract_inverted_index.over	165
abstract_inverted_index.show	149
abstract_inverted_index.that	24, 150
abstract_inverted_index.this	56
abstract_inverted_index.user	28, 43, 79, 133, 138
abstract_inverted_index.with	42
abstract_inverted_index.AILO,	135
abstract_inverted_index.CRA's	154
abstract_inverted_index.ECPO,	67, 128
abstract_inverted_index.Large	3
abstract_inverted_index.These	90
abstract_inverted_index.model	75
abstract_inverted_index.novel	61
abstract_inverted_index.often	20
abstract_inverted_index.these	18
abstract_inverted_index.which	68
abstract_inverted_index.while	118
abstract_inverted_index.(LLMs)	6
abstract_inverted_index.(MTPO)	65
abstract_inverted_index.Agents	15
abstract_inverted_index.Models	5
abstract_inverted_index.Recent	0
abstract_inverted_index.Theory	72
abstract_inverted_index.agents	19
abstract_inverted_index.causes	87, 91
abstract_inverted_index.costly	47
abstract_inverted_index.drives	123
abstract_inverted_index.during	144
abstract_inverted_index.poorly	50
abstract_inverted_index.proven	37
abstract_inverted_index.(CRAs).	16
abstract_inverted_index.address	55
abstract_inverted_index.methods	117
abstract_inverted_index.notable	158
abstract_inverted_index.perform	141
abstract_inverted_index.process	122
abstract_inverted_index.remains	46
abstract_inverted_index.results	148
abstract_inverted_index.support	96, 127
abstract_inverted_index.sustain	27
abstract_inverted_index.thereby	102
abstract_inverted_index.Although	33
abstract_inverted_index.However,	17
abstract_inverted_index.Language	4
abstract_inverted_index.aligning	40
abstract_inverted_index.enhances	153
abstract_inverted_index.ensuring	119
abstract_inverted_index.existing	115, 166
abstract_inverted_index.feedback	139
abstract_inverted_index.generate	21
abstract_inverted_index.guidance	29
abstract_inverted_index.methods.	168
abstract_inverted_index.overhead	113
abstract_inverted_index.paradigm	66
abstract_inverted_index.performs	49
abstract_inverted_index.sampling	112
abstract_inverted_index.simulate	137
abstract_inverted_index.targeted	97
abstract_inverted_index.utilized	94
abstract_inverted_index.LLM-based	132
abstract_inverted_index.achieving	103
abstract_inverted_index.dialogue.	53
abstract_inverted_index.effective	38
abstract_inverted_index.evolution	77
abstract_inverted_index.introduce	59, 130
abstract_inverted_index.leverages	69
abstract_inverted_index.propelled	9
abstract_inverted_index.responses	23
abstract_inverted_index.challenge,	57
abstract_inverted_index.delivering	157
abstract_inverted_index.dialogues,	83
abstract_inverted_index.efficiency	162
abstract_inverted_index.eliminates	109
abstract_inverted_index.explicitly	74
abstract_inverted_index.meaningful	124
abstract_inverted_index.multi-turn	52, 62, 82
abstract_inverted_index.preference	34, 63, 105
abstract_inverted_index.responses,	101
abstract_inverted_index.simulator,	134
abstract_inverted_index.throughout	81
abstract_inverted_index.turn-level	104
abstract_inverted_index.uncovering	84
abstract_inverted_index.underlying	86
abstract_inverted_index.Expectation	70
abstract_inverted_index.development	11
abstract_inverted_index.expectation	142
abstract_inverted_index.ingeniously	108
abstract_inverted_index.interaction	155
abstract_inverted_index.significant	111
abstract_inverted_index.Confirmation	71
abstract_inverted_index.Experimental	147
abstract_inverted_index.advancements	1
abstract_inverted_index.confirmation	143
abstract_inverted_index.improvements	159
abstract_inverted_index.optimization	35, 64, 98, 121
abstract_inverted_index.satisfaction	80
abstract_inverted_index.capabilities,	156
abstract_inverted_index.effectiveness	164
abstract_inverted_index.expectations,	44
abstract_inverted_index.expectations.	32
abstract_inverted_index.improvements.	125
abstract_inverted_index.optimization.	106
abstract_inverted_index.short-sighted	22
abstract_inverted_index.significantly	8, 152
abstract_inverted_index.Conversational	13
abstract_inverted_index.Recommendation	14
abstract_inverted_index.conversational	145
abstract_inverted_index.unsatisfactory	100
abstract_inverted_index.dissatisfaction.	89
abstract_inverted_index.recommendations.	146
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	9
citation_normalized_percentile