A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models

Oleg Silcenco; Marcos R. Machado; Wallace C. Ugulino; Daniel Braun

A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models

Oleg Silcenco
, Marcos R. Machado^*
, Wallace C. Ugulino
, Daniel Braun

^*Corresponding author for this work

Research output: Contribution to conference › Paper › peer-review

Abstract

Aspect-based sentiment analysis enhances sentiment detection by associating it with specific aspects, offering deeper insights than traditional sentiment analysis. This study introduces a manually annotated dataset of 10,814 multilingual customer reviews covering brick-and-mortar retail stores, labeled with eight aspect categories and their sentiment. Using this dataset, the performance of GPT-4 and LLaMA-3 in aspect based sentiment analysis is evaluated to establish a baseline for the newly introduced data. The results show both models achieving over 85% accuracy, while GPT-4 out-performs LLaMA-3 overall with regard to all relevant metrics.

Original language	English
Pages	182-195
Number of pages	14
Publication status	Published - Aug 2025
Event	8th International Conference on Natural Language and Speech Processing, ICNLSP 2025 - University of Southern Denmark, Odense, Denmark Duration: 25 Aug 2025 → 27 Aug 2025 Conference number: 8 https://www.icnlsp.org/2025welcome/

Conference

Conference	8th International Conference on Natural Language and Speech Processing, ICNLSP 2025
Abbreviated title	ICNLSP 2025
Country/Territory	Denmark
City	Odense
Period	25/08/25 → 27/08/25
Internet address	https://www.icnlsp.org/2025welcome/

Access to Document

https://aclanthology.org/2025.icnlsp-1.19.pdfLicence: CC BY

A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models
Silcenco, O., Machad, M. R., Ugulino, W. C. & Braun, D., 25 Aug 2025, ArXiv.org.
Research output: Working paper › Preprint › Academic

Open Access
File
2 Downloads (Pure)

Cite this

@conference{9b0de347474e4ca29d1cd400db30c2bd,

title = "A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models",

abstract = "Aspect-based sentiment analysis enhances sentiment detection by associating it with specific aspects, offering deeper insights than traditional sentiment analysis. This study introduces a manually annotated dataset of 10,814 multilingual customer reviews covering brick-and-mortar retail stores, labeled with eight aspect categories and their sentiment. Using this dataset, the performance of GPT-4 and LLaMA-3 in aspect based sentiment analysis is evaluated to establish a baseline for the newly introduced data. The results show both models achieving over 85\% accuracy, while GPT-4 out-performs LLaMA-3 overall with regard to all relevant metrics.",

author = "Oleg Silcenco and Machado, \{Marcos R.\} and Ugulino, \{Wallace C.\} and Daniel Braun",

year = "2025",

month = aug,

language = "English",

pages = "182--195",

note = "8th International Conference on Natural Language and Speech Processing, ICNLSP 2025, ICNLSP 2025 ; Conference date: 25-08-2025 Through 27-08-2025",

url = "https://www.icnlsp.org/2025welcome/",

}

TY - CONF

T1 - A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models

AU - Silcenco, Oleg

AU - Machado, Marcos R.

AU - Ugulino, Wallace C.

AU - Braun, Daniel

N1 - Conference code: 8

PY - 2025/8

Y1 - 2025/8

N2 - Aspect-based sentiment analysis enhances sentiment detection by associating it with specific aspects, offering deeper insights than traditional sentiment analysis. This study introduces a manually annotated dataset of 10,814 multilingual customer reviews covering brick-and-mortar retail stores, labeled with eight aspect categories and their sentiment. Using this dataset, the performance of GPT-4 and LLaMA-3 in aspect based sentiment analysis is evaluated to establish a baseline for the newly introduced data. The results show both models achieving over 85% accuracy, while GPT-4 out-performs LLaMA-3 overall with regard to all relevant metrics.

AB - Aspect-based sentiment analysis enhances sentiment detection by associating it with specific aspects, offering deeper insights than traditional sentiment analysis. This study introduces a manually annotated dataset of 10,814 multilingual customer reviews covering brick-and-mortar retail stores, labeled with eight aspect categories and their sentiment. Using this dataset, the performance of GPT-4 and LLaMA-3 in aspect based sentiment analysis is evaluated to establish a baseline for the newly introduced data. The results show both models achieving over 85% accuracy, while GPT-4 out-performs LLaMA-3 overall with regard to all relevant metrics.

M3 - Paper

SP - 182

EP - 195

T2 - 8th International Conference on Natural Language and Speech Processing, ICNLSP 2025

Y2 - 25 August 2025 through 27 August 2025

ER -

A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models

Abstract

Conference

Access to Document

Fingerprint

Research output

A Retail-Corpus for Aspect-Based Sentiment Analysis with Large Language Models

Cite this