Taming LLMs with Negative Samples: A Reference-Free Framework to Evaluate Presentation Content with Actionable Feedback Article Swipe

PDF

Ananth Muppidi , Tarak Das , Sambaran Bandyopadhyay , Tripti Shukla , D A Dharun ·

YOU? · · 2025 · Open Access · · DOI: https://doi.org/10.48550/arxiv.2505.18240

The generation of presentation slides automatically is an important problem in the era of generative AI. This paper focuses on evaluating multimodal content in presentation slides that can effectively summarize a document and convey concepts to a broad audience. We introduce a benchmark dataset, RefSlides, consisting of human-made high-quality presentations that span various topics. Next, we propose a set of metrics to characterize different intrinsic properties of the content of a presentation and present REFLEX, an evaluation approach that generates scores and actionable feedback for these metrics. We achieve this by generating negative presentation samples with different degrees of metric-specific perturbations and use them to fine-tune LLMs. This reference-free evaluation technique does not require ground truth presentations during inference. Our extensive automated and human experiments demonstrate that our evaluation approach outperforms classical heuristic-based and state-of-the-art large language model-based evaluations in generating scores and explanations.

Related Topics

The Fragrant Flower Blooms With Dignity

The Girl With The Dragon Tattoo (2011 Film)

My Dinner With Andre

Don't Toy With Me, Miss Nagatoro

List Of Wii Games With Traditional Control Schemes

Cloudy With A Chance Of Meatballs 2

The World Ends With You

Jacob Wrestling With The Angel

Stuck In The Middle With You

Sublime With Rome

The War With Grandpa

Tying The Knot With An Amagami Sister

Concepts

No concepts available.

Metadata

Type: preprint
Language: en
Landing Page: http://arxiv.org/abs/2505.18240
PDF: https://arxiv.org/pdf/2505.18240
OA Status: green
OpenAlex ID: https://openalex.org/W4414581222

All OpenAlex metadata

Raw OpenAlex JSON

OpenAlex ID: https://openalex.org/W4414581222

Canonical identifier for this work in OpenAlex
DOI: https://doi.org/10.48550/arxiv.2505.18240

Digital Object Identifier
Title: Taming LLMs with Negative Samples: A Reference-Free Framework to Evaluate Presentation Content with Actionable Feedback

Work title
Type: preprint

OpenAlex work type
Language: en

Primary language
Publication year: 2025

Year of publication
Publication date: 2025-05-23

Full publication date if available
Authors: Ananth Muppidi, Tarak Das, Sambaran Bandyopadhyay, Tripti Shukla, D A Dharun

List of authors in order
Landing page: https://arxiv.org/abs/2505.18240

Publisher landing page
PDF URL: https://arxiv.org/pdf/2505.18240

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/2505.18240

Direct OA link when available
Cited by: 0

Total citation count in OpenAlex

Full payload

id	https://openalex.org/W4414581222
doi	https://doi.org/10.48550/arxiv.2505.18240
ids.doi	https://doi.org/10.48550/arxiv.2505.18240
ids.openalex	https://openalex.org/W4414581222
fwci
type	preprint
title	Taming LLMs with Negative Samples: A Reference-Free Framework to Evaluate Presentation Content with Actionable Feedback
biblio.issue
biblio.volume
biblio.last_page
biblio.first_page
topics[0].id	https://openalex.org/T10181
topics[0].field.id	https://openalex.org/fields/17
topics[0].field.display_name	Computer Science
topics[0].score	0.9208999872207642
topics[0].domain.id	https://openalex.org/domains/3
topics[0].domain.display_name	Physical Sciences
topics[0].subfield.id	https://openalex.org/subfields/1702
topics[0].subfield.display_name	Artificial Intelligence
topics[0].display_name	Natural Language Processing Techniques
topics[1].id	https://openalex.org/T13643
topics[1].field.id	https://openalex.org/fields/33
topics[1].field.display_name	Social Sciences
topics[1].score	0.9096999764442444
topics[1].domain.id	https://openalex.org/domains/2
topics[1].domain.display_name	Social Sciences
topics[1].subfield.id	https://openalex.org/subfields/3320
topics[1].subfield.display_name	Political Science and International Relations
topics[1].display_name	Artificial Intelligence in Law
is_xpac	False
apc_list
apc_paid
language	en
locations[0].id	pmh:oai:arXiv.org:2505.18240
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/2505.18240
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/2505.18240
locations[1].id	doi:10.48550/arxiv.2505.18240
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.2505.18240
indexed_in	arxiv, datacite
authorships[0].author.id	https://openalex.org/A5119181797
authorships[0].author.orcid
authorships[0].author.display_name	Ananth Muppidi
authorships[0].author_position	first
authorships[0].raw_author_name	Muppidi, Ananth
authorships[0].is_corresponding	False
authorships[1].author.id	https://openalex.org/A5072004888
authorships[1].author.orcid
authorships[1].author.display_name	Tarak Das
authorships[1].author_position	middle
authorships[1].raw_author_name	Das, Tarak
authorships[1].is_corresponding	False
authorships[2].author.id	https://openalex.org/A5111227747
authorships[2].author.orcid
authorships[2].author.display_name	Sambaran Bandyopadhyay
authorships[2].author_position	middle
authorships[2].raw_author_name	Bandyopadhyay, Sambaran
authorships[2].is_corresponding	False
authorships[3].author.id	https://openalex.org/A5013402151
authorships[3].author.orcid	https://orcid.org/0000-0001-8250-2431
authorships[3].author.display_name	Tripti Shukla
authorships[3].author_position	middle
authorships[3].raw_author_name	Shukla, Tripti
authorships[3].is_corresponding	False
authorships[4].author.id	https://openalex.org/A5119753504
authorships[4].author.orcid
authorships[4].author.display_name	D A Dharun
authorships[4].author_position	last
authorships[4].raw_author_name	A, Dharun D
authorships[4].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/2505.18240
open_access.oa_status	green
open_access.any_repository_has_fulltext	False
created_date	2025-10-10T00:00:00
display_name	Taming LLMs with Negative Samples: A Reference-Free Framework to Evaluate Presentation Content with Actionable Feedback
has_fulltext	False
is_retracted	False
updated_date	2025-11-06T06:51:31.235846
primary_topic.id	https://openalex.org/T10181
primary_topic.field.id	https://openalex.org/fields/17
primary_topic.field.display_name	Computer Science
primary_topic.score	0.9208999872207642
primary_topic.domain.id	https://openalex.org/domains/3
primary_topic.domain.display_name	Physical Sciences
primary_topic.subfield.id	https://openalex.org/subfields/1702
primary_topic.subfield.display_name	Artificial Intelligence
primary_topic.display_name	Natural Language Processing Techniques
cited_by_count	0
locations_count	2
best_oa_location.id	pmh:oai:arXiv.org:2505.18240
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/2505.18240
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/2505.18240
primary_location.id	pmh:oai:arXiv.org:2505.18240
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/2505.18240
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/2505.18240
publication_date	2025-05-23
publication_year	2025
referenced_works_count	0
abstract_inverted_index.a	30, 36, 41, 57, 70
abstract_inverted_index.We	39, 87
abstract_inverted_index.an	7, 75
abstract_inverted_index.by	90
abstract_inverted_index.in	10, 23, 139
abstract_inverted_index.is	6
abstract_inverted_index.of	2, 13, 46, 59, 66, 69, 98
abstract_inverted_index.on	19
abstract_inverted_index.to	35, 61, 104
abstract_inverted_index.we	55
abstract_inverted_index.AI.	15
abstract_inverted_index.Our	119
abstract_inverted_index.The	0
abstract_inverted_index.and	32, 72, 81, 101, 122, 133, 142
abstract_inverted_index.can	27
abstract_inverted_index.era	12
abstract_inverted_index.for	84
abstract_inverted_index.not	112
abstract_inverted_index.our	127
abstract_inverted_index.set	58
abstract_inverted_index.the	11, 67
abstract_inverted_index.use	102
abstract_inverted_index.This	16, 107
abstract_inverted_index.does	111
abstract_inverted_index.span	51
abstract_inverted_index.that	26, 50, 78, 126
abstract_inverted_index.them	103
abstract_inverted_index.this	89
abstract_inverted_index.with	95
abstract_inverted_index.LLMs.	106
abstract_inverted_index.Next,	54
abstract_inverted_index.broad	37
abstract_inverted_index.human	123
abstract_inverted_index.large	135
abstract_inverted_index.paper	17
abstract_inverted_index.these	85
abstract_inverted_index.truth	115
abstract_inverted_index.convey	33
abstract_inverted_index.during	117
abstract_inverted_index.ground	114
abstract_inverted_index.scores	80, 141
abstract_inverted_index.slides	4, 25
abstract_inverted_index.REFLEX,	74
abstract_inverted_index.achieve	88
abstract_inverted_index.content	22, 68
abstract_inverted_index.degrees	97
abstract_inverted_index.focuses	18
abstract_inverted_index.metrics	60
abstract_inverted_index.present	73
abstract_inverted_index.problem	9
abstract_inverted_index.propose	56
abstract_inverted_index.require	113
abstract_inverted_index.samples	94
abstract_inverted_index.topics.	53
abstract_inverted_index.various	52
abstract_inverted_index.approach	77, 129
abstract_inverted_index.concepts	34
abstract_inverted_index.dataset,	43
abstract_inverted_index.document	31
abstract_inverted_index.feedback	83
abstract_inverted_index.language	136
abstract_inverted_index.metrics.	86
abstract_inverted_index.negative	92
abstract_inverted_index.audience.	38
abstract_inverted_index.automated	121
abstract_inverted_index.benchmark	42
abstract_inverted_index.classical	131
abstract_inverted_index.different	63, 96
abstract_inverted_index.extensive	120
abstract_inverted_index.fine-tune	105
abstract_inverted_index.generates	79
abstract_inverted_index.important	8
abstract_inverted_index.intrinsic	64
abstract_inverted_index.introduce	40
abstract_inverted_index.summarize	29
abstract_inverted_index.technique	110
abstract_inverted_index.RefSlides,	44
abstract_inverted_index.actionable	82
abstract_inverted_index.consisting	45
abstract_inverted_index.evaluating	20
abstract_inverted_index.evaluation	76, 109, 128
abstract_inverted_index.generating	91, 140
abstract_inverted_index.generation	1
abstract_inverted_index.generative	14
abstract_inverted_index.human-made	47
abstract_inverted_index.inference.	118
abstract_inverted_index.multimodal	21
abstract_inverted_index.properties	65
abstract_inverted_index.demonstrate	125
abstract_inverted_index.effectively	28
abstract_inverted_index.evaluations	138
abstract_inverted_index.experiments	124
abstract_inverted_index.model-based	137
abstract_inverted_index.outperforms	130
abstract_inverted_index.characterize	62
abstract_inverted_index.high-quality	48
abstract_inverted_index.presentation	3, 24, 71, 93
abstract_inverted_index.automatically	5
abstract_inverted_index.explanations.	143
abstract_inverted_index.perturbations	100
abstract_inverted_index.presentations	49, 116
abstract_inverted_index.reference-free	108
abstract_inverted_index.heuristic-based	132
abstract_inverted_index.metric-specific	99
abstract_inverted_index.state-of-the-art	134
cited_by_percentile_year
countries_distinct_count	0
institutions_distinct_count	5
citation_normalized_percentile