Embedding generation for text classification of Brazilian Portuguese user reviews: from bag-of-words to transformers

Souza, Frederico Dias; Filho, João Baptista de Oliveira e Souza

doi:10.1007/s00521-022-08068-6

Cited by 4 publications

(4 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…O transformer BERT, arquitetado em Python, foi utilizado na etapa de Pós-Processamento das informações extraídas (SOUZA; FILHO, 2023). Forneceu a descoberta de palavraschave mais recorrentes nas classificações de extratos do Corpus de análise em Funções de Combate.…”

Section: O Processamento De Informaçõesunclassified

O processamento de informações de ações militares da MONUSCO: uma proposta de utilização de inteligência artificial para a Prospecção Tecnológica.

CUNHA,

Passos,

Salles

2024

Anais Do(a) Anais Do Congresso Brasileiro De Inovação E Gestão De Desenvolvimento Do Produto

View full text Add to dashboard Cite

The present study was conducted as a Knowledge Discovery in Databases (KDD) approach with the objective of carrying out Text Mining on MONUSCO reports to identify information for the benefit of Technology Foresight in that mission. The method used followed the KDD operational stages of Pre-Processing, Data Mining (Text Mining) and Post-Processing, using a Recurrent Neural Network and BERT transformer, to fulfill the textual classification tasks. As main results, the feasibility of using artificial intelligence for this purpose was observed, having obtained a satisfactory classification in the Combat Functions: Command and Control, Movement and Maneuver, Intelligence, Fires, Logistics and Protection. Furthermore, the usefulness of this aspect of processing textual data, obtained from a corpus made up of reports, was verified as a rich source of useful information for associating the results obtained with technological branches defined in a specific taxonomy. Finally, the present work proved to be a viable modality for carrying out Technology Foresight in areas of conflict with Text Mining, adopting Artificial Intelligence resources to meet the demands of Natural Language Processing.

show abstract

Section: O Processamento De Informaçõesunclassified

O processamento de informações de ações militares da MONUSCO: uma proposta de utilização de inteligência artificial para a Prospecção Tecnológica.

CUNHA,

Passos,

Salles

2024

Anais Do(a) Anais Do Congresso Brasileiro De Inovação E Gestão De Desenvolvimento Do Produto

View full text Add to dashboard Cite

show abstract

“…In addition, articles demonstrate that transformer-based models achieve superior results compared to previous models, such as the case of LSTM and CNN [24]. It is possible to adjust the hyperparameters in the training of the BERT and RoBERTa models to achieve better results.…”

Section: Related Workmentioning

confidence: 99%

Sentiment Analysis in Portuguese Restaurant Reviews: Application of Transformer Models in Edge Computing

Branco,

Parada,

Silva

et al. 2024

Electronics

View full text Add to dashboard Cite

This study focuses on improving sentiment analysis in restaurant reviews by leveraging transfer learning and transformer-based pre-trained models. This work evaluates the suitability of pre-trained deep learning models for analyzing Natural Language Processing tasks in Portuguese. It also explores the viability of utilizing edge devices for Natural Language Processing tasks, considering their computational limitations and resource constraints. Specifically, we employ bidirectional encoder representations from transformers and robustly optimized BERT approach, two state-of-the-art models, to build a sentiment review classifier. The classifier’s performance is evaluated using accuracy and area under the receiver operating characteristic curve as the primary metrics. Our results demonstrate that the classifier developed using ensemble techniques outperforms the baseline model (from 0.80 to 0.84) in accurately classifying restaurant review sentiments when three classes are considered (negative, neutral, and positive), reaching an accuracy and area under the receiver operating characteristic curve higher than 0.8 when examining a Zomato restaurant review dataset, provided for this work. This study seeks to create a model for the precise classification of Portuguese reviews into positive, negative, or neutral categories. The flexibility of deploying our model on affordable hardware platforms suggests its potential to enable real-time solutions. The deployment of the model on edge computing platforms improves accessibility in resource-constrained environments.

show abstract

“…It produces embeddings that ignore the order, context, and semantic relations between words in a corpus. It encompasses statistics extracted from the corpus, such as the word frequency or the term Frequency-Inverse Document Frequency (TF-IDF) (12). Bag-of-words is one the most common text representation method for natural language processing.…”

Section: Bag-of-words Modelmentioning

confidence: 99%

Performance evaluation of random forest algorithm for automating classification of mathematics question items

Baladjay¹,

Riva²,

Santos³

et al. 2023

World J. Adv. Res. Rev.

View full text Add to dashboard Cite

Automated classification of mathematics question items based on the Table of Specifications is crucial in developing well-defined assessment content, significantly reducing teachers’ workload. This study presents a performance evaluation of a Random Forest model designed to classify mathematics question items based on the content standards of the first quarter of tenth grade stipulated by the Philippines’ Department of Education Curriculum Guide. The model uses an algorithm that extracts mathematical expressions as tokens for the Bag-of-words Model. The evaluation was conducted using precision, recall, F-1 score, and overall accuracy metrics, and the confusion matrix was used to assess the Random Forest model’s performance. The results showed that the Random Forest model achieved 95% in precision, 95% in recall, 95% in F-1 score, and 95% in overall accuracy, demonstrating its effectiveness in classifying mathematics question items.

show abstract

Embedding generation for text classification of Brazilian Portuguese user reviews: from bag-of-words to transformers

Cited by 4 publications

References 21 publications

O processamento de informações de ações militares da MONUSCO: uma proposta de utilização de inteligência artificial para a Prospecção Tecnológica.

O processamento de informações de ações militares da MONUSCO: uma proposta de utilização de inteligência artificial para a Prospecção Tecnológica.

Sentiment Analysis in Portuguese Restaurant Reviews: Application of Transformer Models in Edge Computing

Performance evaluation of random forest algorithm for automating classification of mathematics question items

Contact Info

Product

Resources

About