Context Attribution with Multi-Armed Bandit Optimization Article Swipe

PDF

Deng Pan , Keerthiram Murugesan , Nuno Moniz , Nitesh V. Chawla ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2506.19977

Understanding which parts of the retrieved context contribute to a large language model's generated answer is essential for building interpretable and trustworthy generative QA systems. We propose a novel framework that formulates context attribution as a combinatorial multi-armed bandit (CMAB) problem. Each context segment is treated as a bandit arm, and we employ Combinatorial Thompson Sampling (CTS) to efficiently explore the exponentially large space of context subsets under a limited query budget. Our method defines a reward function based on normalized token likelihoods, capturing how well a subset of segments supports the original model response. Unlike traditional perturbation-based attribution methods such as SHAP, which sample subsets uniformly and incur high computational costs, our approach adaptively balances exploration and exploitation by leveraging posterior estimates of segment relevance. This leads to substantially improved query efficiency while maintaining high attribution fidelity. Extensive experiments on diverse datasets and LLMs demonstrate that our method achieves competitive attribution quality with fewer model queries.

Related Topics

List Of Nba Players With Most Championships

Married... With Children

Anne With An E

Just Go With It

The Man With The Golden Gun (Film)

Friends With Benefits (Film)

The Girl With The Dragon Tattoo

Gone With The Wind (Novel)

The Hero With A Thousand Faces

Sleeping With The Enemy

List Of Men's Footballers With 50 Or More International Goals

It Is Well With My Soul

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2506.19977
PDF: https://arxiv.org/pdf/2506.19977
OA Status: green
OpenAlex ID: https://openalex.org/W4414988062

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4414988062

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2506.19977

Digital Object Identifier
Title: Context Attribution with Multi-Armed Bandit Optimization

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-06-24

Full publication date if available
Authors: Deng Pan, Keerthiram Murugesan, Nuno Moniz, Nitesh V. Chawla

List of authors in order
Landing page: https://arxiv.org/abs/2506.19977

Publisher landing page
PDF URL: https://arxiv.org/pdf/2506.19977

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/2506.19977

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4414988062
doi	https://doi.org/10.48550/arxiv.2506.19977
ids.doi	https://doi.org/10.48550/arxiv.2506.19977
ids.openalex	https://openalex.org/W4414988062
fwci
type	preprint
title	Context Attribution with Multi-Armed Bandit Optimization
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T12101
topics[0].field.id	https://openalex.org/fields/18
topics[0].field.display_name	Decision Sciences
topics[0].score	0.9936000108718872
topics[0].domain.id	https://openalex.org/domains/2
topics[0].domain.display_name	Social Sciences
topics[0].subfield.id	https://openalex.org/subfields/1803
topics[0].subfield.display_name	Management Science and Operations Research
topics[0].display_name	Advanced Bandit Algorithms Research
topics[1].id	https://openalex.org/T10203
topics[1].field.id	https://openalex.org/fields/17
topics[1].field.display_name	Computer Science
topics[1].score	0.9598000049591064
topics[1].domain.id	https://openalex.org/domains/3
topics[1].domain.display_name	Physical Sciences
topics[1].subfield.id	https://openalex.org/subfields/1710
topics[1].subfield.display_name	Information Systems
topics[1].display_name	Recommender Systems and Techniques
topics[2].id	https://openalex.org/T12761
topics[2].field.id	https://openalex.org/fields/17
topics[2].field.display_name	Computer Science
topics[2].score	0.9484000205993652
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	Data Stream Mining Techniques
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2506.19977
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/2506.19977
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/2506.19977
locations[1].id	doi:10.48550/arxiv.2506.19977
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	cc-by
locations[1].pdf_url
locations[1].version
locations[1].raw_type	article
locations[1].license_id	https://openalex.org/licenses/cc-by
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.2506.19977
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5103197872
authorships[0].author.orcid	https://orcid.org/0000-0002-1302-9182
authorships[0].author.display_name	Deng Pan
authorships[0].author_position	first
authorships[0].raw_author_name	Pan, Deng
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5104188415
authorships[1].author.orcid
authorships[1].author.display_name	Keerthiram Murugesan
authorships[1].author_position	middle
authorships[1].raw_author_name	Murugesan, Keerthiram
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5047689411
authorships[2].author.orcid	https://orcid.org/0000-0003-4322-1076
authorships[2].author.display_name	Nuno Moniz
authorships[2].author_position	middle
authorships[2].raw_author_name	Moniz, Nuno
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5068157871
authorships[3].author.orcid	https://orcid.org/0000-0003-3932-5956
authorships[3].author.display_name	Nitesh V. Chawla
authorships[3].author_position	last
authorships[3].raw_author_name	Chawla, Nitesh
authorships[3].is_corresponding	False
has_content.pdf	False
has_content.grobid_xml	False
is_paratext	False
open_access.is_oa	True
open_access.oa_url	https://arxiv.org/pdf/2506.19977
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-09T00:00:00
display_name	Context Attribution with Multi-Armed Bandit Optimization
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T12101
primary_topic.field.id	https://openalex.org/fields/18
primary_topic.field.display_name	Decision Sciences
primary_topic.score	0.9936000108718872
primary_topic.domain.id	https://openalex.org/domains/2
primary_topic.domain.display_name	Social Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1803
primary_topic.subfield.display_name	Management Science and Operations Research
primary_topic.display_name	Advanced Bandit Algorithms Research
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2506.19977
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/2506.19977
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/2506.19977
primary_location.id	pmh:oai:arXiv.org:2506.19977
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/2506.19977
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/2506.19977
publication_date	2025-06-24
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	9, 27, 35, 47, 68, 75, 86
abstract_inverted_index.QA	23
abstract_inverted_index.We	25
abstract_inverted_index.as	34, 46, 101
abstract_inverted_index.by	119
abstract_inverted_index.is	15, 44
abstract_inverted_index.of	3, 64, 88, 123
abstract_inverted_index.on	79, 140
abstract_inverted_index.to	8, 57, 128
abstract_inverted_index.we	51
abstract_inverted_index.Our	72
abstract_inverted_index.and	20, 50, 107, 117, 143
abstract_inverted_index.for	17
abstract_inverted_index.how	84
abstract_inverted_index.our	112, 147
abstract_inverted_index.the	4, 60, 91
abstract_inverted_index.Each	41
abstract_inverted_index.LLMs	144
abstract_inverted_index.This	126
abstract_inverted_index.arm,	49
abstract_inverted_index.high	109, 135
abstract_inverted_index.such	100
abstract_inverted_index.that	30, 146
abstract_inverted_index.well	85
abstract_inverted_index.with	153
abstract_inverted_index.(CTS)	56
abstract_inverted_index.SHAP,	102
abstract_inverted_index.based	78
abstract_inverted_index.fewer	154
abstract_inverted_index.incur	108
abstract_inverted_index.large	10, 62
abstract_inverted_index.leads	127
abstract_inverted_index.model	93, 155
abstract_inverted_index.novel	28
abstract_inverted_index.parts	2
abstract_inverted_index.query	70, 131
abstract_inverted_index.space	63
abstract_inverted_index.token	81
abstract_inverted_index.under	67
abstract_inverted_index.which	1, 103
abstract_inverted_index.while	133
abstract_inverted_index.(CMAB)	39
abstract_inverted_index.Unlike	95
abstract_inverted_index.answer	14
abstract_inverted_index.bandit	38, 48
abstract_inverted_index.costs,	111
abstract_inverted_index.employ	52
abstract_inverted_index.method	73, 148
abstract_inverted_index.reward	76
abstract_inverted_index.sample	104
abstract_inverted_index.subset	87
abstract_inverted_index.budget.	71
abstract_inverted_index.context	6, 32, 42, 65
abstract_inverted_index.defines	74
abstract_inverted_index.diverse	141
abstract_inverted_index.explore	59
abstract_inverted_index.limited	69
abstract_inverted_index.methods	99
abstract_inverted_index.model's	12
abstract_inverted_index.propose	26
abstract_inverted_index.quality	152
abstract_inverted_index.segment	43, 124
abstract_inverted_index.subsets	66, 105
abstract_inverted_index.treated	45
abstract_inverted_index.Sampling	55
abstract_inverted_index.Thompson	54
abstract_inverted_index.achieves	149
abstract_inverted_index.approach	113
abstract_inverted_index.balances	115
abstract_inverted_index.building	18
abstract_inverted_index.datasets	142
abstract_inverted_index.function	77
abstract_inverted_index.improved	130
abstract_inverted_index.language	11
abstract_inverted_index.original	92
abstract_inverted_index.problem.	40
abstract_inverted_index.queries.	156
abstract_inverted_index.segments	89
abstract_inverted_index.supports	90
abstract_inverted_index.systems.	24
abstract_inverted_index.Extensive	138
abstract_inverted_index.capturing	83
abstract_inverted_index.essential	16
abstract_inverted_index.estimates	122
abstract_inverted_index.fidelity.	137
abstract_inverted_index.framework	29
abstract_inverted_index.generated	13
abstract_inverted_index.posterior	121
abstract_inverted_index.response.	94
abstract_inverted_index.retrieved	5
abstract_inverted_index.uniformly	106
abstract_inverted_index.adaptively	114
abstract_inverted_index.contribute	7
abstract_inverted_index.efficiency	132
abstract_inverted_index.formulates	31
abstract_inverted_index.generative	22
abstract_inverted_index.leveraging	120
abstract_inverted_index.normalized	80
abstract_inverted_index.relevance.	125
abstract_inverted_index.attribution	33, 98, 136, 151
abstract_inverted_index.competitive	150
abstract_inverted_index.demonstrate	145
abstract_inverted_index.efficiently	58
abstract_inverted_index.experiments	139
abstract_inverted_index.exploration	116
abstract_inverted_index.maintaining	134
abstract_inverted_index.multi-armed	37
abstract_inverted_index.traditional	96
abstract_inverted_index.trustworthy	21
abstract_inverted_index.exploitation	118
abstract_inverted_index.likelihoods,	82
abstract_inverted_index.Combinatorial	53
abstract_inverted_index.Understanding	0
abstract_inverted_index.combinatorial	36
abstract_inverted_index.computational	110
abstract_inverted_index.exponentially	61
abstract_inverted_index.interpretable	19
abstract_inverted_index.substantially	129
abstract_inverted_index.perturbation-based	97
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	4
citation_normalized_percentile