Computational Processing of the Portuguese Language

Preface
Organization
Contents
Resources and Evaluation
UlyssesNER-Br: A Corpus of Brazilian Legislative Documents for Named Entity Recognition
	1 Introduction
	2 Related Work
	3 Method
		3.1 Semantic Classes
		3.2 Annotation Process
	4 The UlyssesNER-Br Corpus
		4.1 PL-corpus
		4.2 ST-corpus
		4.3 Evaluation
		4.4 Results and Discussion
	5 Conclusion and Future Works
	References
A Test Suite for the Evaluation of Portuguese-English Machine Translation
	1 Introduction
	2 Related Work
	3 Method
		3.1 Creation of the Test Suite
		3.2 Limitations of the Method
		3.3 Experimental Setup
	4 Findings
		4.1 Overall Performance of MT Systems
		4.2 BLEU vs. Test Suite Scores
		4.3 Categories
		4.4 Phenomena
		4.5 Qualitative Analysis
	5 Conclusion
	References
MINT - Mainstream and Independent News Text Corpus
	1 Introduction
	2 Related Work
	3 Corpus Organization
		3.1 MINT-articles
		3.2 MINT-annotations
	4 Corpus Characterization
		4.1 Linguistic Characterization
		4.2 Insights from Crowdsourced Annotations
	5 Conclusion
	References
Fakepedia Corpus: A Flexible Fake News Corpus in Portuguese
	1 Introduction
	2 Related Work
	3 Our Proposal: Fakepedia Corpus
	4 Experiments and Results
	5 Conclusion
	References
A Targeted Assessment of the Syntactic Abilities of Transformer Models for Galician-Portuguese
	1 Introduction
	2 Related Work
	3 Materials and Methods
		3.1 Experiments and Data
		3.2 Models
		3.3 Evaluation
	4 Results and Discussion
	5 Conclusions and Further Work
	References
FakeRecogna: A New Brazilian Corpus for Fake News Detection
	1 Introduction
	2 Related Works
	3 FakeRecogna Corpus
	4 Methodology
		4.1 Pre-processing
		4.2 Text Representation
		4.3 Classifiers
		4.4 Evaluating Measures
		4.5 Additional Experiments
	5 Experimental Results
		5.1 No Removal of Words
		5.2 Augmentation Study
	6 Conclusions and Future Works
	References
Implicit Opinion Aspect Clues in Portuguese Texts: Analysis and Categorization
	1 Introduction
	2 Related Work
	3 Data and Methods
		3.1 Methods
		3.2 Datasets
		3.3 Identification of IACs
		3.4 Lexicons of IACs
		3.5 Categorization of IACs
	4 Results
	5 Final Remarks
	References
CRPC-DB a Discourse Bank for Portuguese
	1 Introduction
	2 Related Work
	3 The CRPC-DB
		3.1 Raw Corpus and Pre-processing
		3.2 Annotation Scheme
		3.3 Annotation Process
	4 Inter-Annotator Agreement Experiment
	5 Final Remarks
	References
Challenges in Annotating a Treebank of Clinical Narratives in Brazilian Portuguese
	1 Introduction
	2 Related Work
	3 Materials and Methods
		3.1 Data Preparation
		3.2 Corpus Characteristics
		3.3 Decisions Made in the Annotation Process
	4 Results
	5 Discussion
	6 Conclusion
	References
PetroBERT: A Domain Adaptation Language Model for Oil and Gas Applications in Portuguese
	1 Introduction
	2 Related Works
	3 Proposed Work
	4 Experimental Evaluation and Discussion
		4.1 Experiment I - NER
		4.2 Experiment II - Sentence Classification
	5 Conclusions and Future Works
	References
SS-PT: A Stance and Sentiment Data Set from Portuguese Quoted Tweets
	1 Introduction
	2 Related Work
	3 Corpus
		3.1 Data Collection
		3.2 Guidelines
		3.3 Annotation Procedure
		3.4 Balancing the Corpus
		3.5 Descriptive Statistics
	4 Baseline Experiment
	5 Conclusion
	References
Natural Language Processing Tasks
ZeroBERTo: Leveraging Zero-Shot Text Classification by Topic Modeling
	1 Introduction
	2 Background and Related Work
	3 Proposed Method
		3.1 0shot-TC Task Formalization
		3.2 ZeroBERTo
	4 Experiments
	5 Discussion and Future Work
	References
Banking Regulation Classification in Portuguese
	1 Introduction
	2 Related Works
	3 The Application
		3.1 The Corpus
		3.2 The Architecture
	4 Discussion, Results and Future Work
	5 Conclusions
	References
Automatic Information Extraction: A Distant Reading of the Brazilian Historical-Biographical Dictionary
	1 Introduction
	2 Information Extraction
	3 Methodology
	4 Extraction Evaluation
	5 Distant Reading DHBB
	6 Final Considerations
	References
Automatic Recognition of Units of Measurement in Product Descriptions from Tax Invoices Using Neural Networks
	1 Introduction
	2 Related Works
	3 Methodology
		3.1 Materials
		3.2 Dataset
		3.3 Data Preparation
		3.4 Training and Test
	4 Results and Discussion
		4.1 General Analysis
		4.2 Analysis of Errors
	5 Conclusions
	References
Entity Extraction from Portuguese Legal Documents Using Distant Supervision
	1 Introduction
	2 Related Work
	3 Entity Extraction System
	4 Experimental Results
		4.1 Dataset and Evaluation Metric
		4.2 Best Model Results
		4.3 DAM: Role-Specific Threshold
		4.4 Assessment of DAM Components
	5 Conclusion
	References
Sexist Hate Speech: Identifying Potential Online Verbal Violence Instances
	1 Introduction
	2 Background
		2.1 The Campos Mello Case
	3 The Linguistic-Computational Interface
	4 Computational Approaches to Support Hate Speech Identification Through Linguistic Characteristics
		4.1 Fallacies in Intolerant Speech
	5 Final Remarks
	References
Book Genre Classification Based on Reviews of Portuguese-Language Literature
	1 Introduction
	2 Related Work
	3 Genre Classification Methodology
		3.1 Data
		3.2 Reviews Preprocessing
		3.3 Book Genre Classification
		3.4 Evaluation
	4 Experiments and Results
	5 Conclusion and Future Work
	References
Combining Word Embeddings for Portuguese Named Entity Recognition
	1 Introduction
	2 Related Work
	3 Resources
		3.1 Corpora
		3.2 Word Embedding Models
		3.3 NER Classification Model
	4 Experiments
		4.1 Experimental Setup
		4.2 Experimental Results
	5 Conclusion
	References
BERT for Sentiment Analysis: Pre-trained and Fine-Tuned Alternatives
	1 Introduction
	2 Related Work
	3 Datasets
	4 Models
		4.1 Pre-trained BERT
		4.2 Fine-Tuned BERT
	5 Results and Discussion
		5.1 Pre-trained BERT
		5.2 Fine-Tuned BERT
		5.3 Cross-model Comparison
	6 Conclusion
	References
Fostering Judiciary Applications with New Fine-Tuned Models for Legal Named Entity Recognition in Portuguese
	1 Introduction
	2 Related Work
	3 Materials and Methods
		3.1 Fine-Tuned Legal NER
		3.2 Prototype Application
	4 Results and Discussion
		4.1 Fine-Tuned Legal NER
		4.2 Prototype Application
	5 Conclusion
	References
Natural Language Processing Applications
Using Topic Modeling in Classification of Brazilian Lawsuits
	1 Introduction
	2 Related Works
	3 Corpus and Data Preparation
		3.1 Corpus and Golden Collection
		3.2 Integration with the Brazilian Legal Knowledge Graph
	4 Topic Modeling in Legal Documents
		4.1 Pre-processing
		4.2 Topic Generation
		4.3 Converting Topics to Feature Vectors
	5 Experiments
		5.1 Experimental Setup
		5.2 Models
		5.3 Experimental Results and Analysis
	6 Conclusions
	References
PortNOIE: A Neural Framework for Open Information Extraction for the Portuguese Language
	1 Introduction
	2 Related Work
	3 PortNOIE
		3.1 Problem Definition
		3.2 Architecture
	4 Experiments
		4.1 Datasets
		4.2 Experimental Design
		4.3 Results
		4.4 Ablation and Discussion
	5 Conclusion and Future Work
	References
Tracking Environmental Policy Changes in the Brazilian Federal Official Gazette
	1 Introduction
	2 Methods
		2.1 Data Preparation
		2.2 Experiment Description
	3 Results
	4 Conclusion and Future Work
	References
A Transfer Learning Analysis of Political Leaning Classification in Cross-domain Content
	1 Introduction
	2 Related Work
	3 Data Collection
	4 Experiments
		4.1 Congressional Speeches Classification
		4.2 Transfer Learning Classification
		4.3 Transfer Learning Decay over Time
	5 Discussion
	6 Limitations and Future Work
	References
Integrating Question Answering and Text-to-SQL in Portuguese
	1 Introduction
	2 Background and Tools
	3 Proposed Architecture
	4 Question Answering Datasets
	5 Experiments
		5.1 Classifier
		5.2 Question Answering Reasoner
	6 Results and Analyses
	7 Conclusion
	References
Named Entity Extractors for New Domains by Transfer Learning with Automatically Annotated Data
	1 Introduction
	2 Related Work
	3 Method
		3.1 Datasets
		3.2 BERT-Based Classifiers for Entity Detection
	4 Results
	5 Conclusion
		5.1 Future Work
	References
PTT5-Paraphraser: Diversity and Meaning Fidelity in Automatic Portuguese Paraphrasing
	1 Introduction
	2 Related Work
	3 The PTT5-Paraphraser
	4 Evaluating Paraphrasers
		4.1 Evaluation by Computational Metrics
		4.2 Human Evaluation
	5 Data Augmentation Experiment
	6 Conclusions
	References
Speech Processing and Applications
A Protocol for Comparing Gesture and Prosodic Boundaries in Multimodal Corpora
	1 Gesture and Prosody Alignment Background
		1.1 The Alignment and Its Types
		1.2 The Language into Act Theory
		1.3 BGEST Corpus Overview
	2 Script Outline
	3 Results and Discussion
	References
Forced Phonetic Alignment in Brazilian Portuguese Using Time-Delay Neural Networks
	1 Introduction
	2 Model Training and Evaluation Tests
		2.1 Evaluation Speech Corpus
		2.2 Simulation Overview
	3 Results and Discussion
	4 Conclusion
	References
Brazilian Portuguese Speech Recognition Using Wav2vec 2.0
	1 Introduction
	2 Wav2vec
	3 ASRs for Brazilian Portuguese
	4 Datasets and Experiments
	5 Results and Discussion
	6 Conclusions
	References
A Corpus of Neutral Voice Speech in Brazilian Portuguese
	1 Introduction
		1.1 Related Works
	2 Objectives
	3 Dataset
		3.1 Text
		3.2 Recordings
		3.3 Data Structure
		3.4 Statistics
		3.5 Data Access
	4 Applications
	5 Experiments
	6 Conclusion
	References
Comparing Lexical and Usage Frequencies of Palatal Segments in Portuguese
	1 Introduction
	2 Methods
	3 Results
		3.1 Context-Free Frequency
		3.2 Word Position Frequency
		3.3 Left Adjacent Vowel Frequency
	4 Discussion
	References
Lexical Semantics
Extracting Valences from a Dependency Treebank for Populating the Verb Lexicon of a Portuguese HPSG Grammar
	1 Introduction
	2 Valence Encoding in PorGram
	3 Universal Dependencies and the UD Bosque Treebank
	4 From Dependency Information to Valence Types
	5 Final Remarks
	References
CQL Grammars for Lexical and Semantic Information Extraction for Portuguese and Italian
	1 Introduction
		1.1 General Methodology
	2 Specialized Lexical Units Extraction: Frequency Analysis Wrt Reference Corpora
		2.1 CORPORART vs Monitor Corpora: Data Analysis and Results
	3 Concordances Analysis Through Lexical-Syntactic Patterns
		3.1 Knowledge Patterns and Sketch Grammar
		3.2 Lexical-Syntactic Patterns Evaluation and Discussion
	4 Final Remarks and Future Work
	References
Drilling Lexico-Semantic Knowledge in Portuguese from BERT
	1 Introduction
	2 Related Work
	3 Experimentation Setup
		3.1 Considerations on the Patterns
		3.2 Dataset
		3.3 Metrics
		3.4 Baselines
	4 Performance with Different Patterns
		4.1 Single Patterns for Non-symmetric Relations
		4.2 Combining the Best Patterns
	5 Conclusions
	References
Short Papers
The Systematic Construction of Multiple Types of Corpora Through the Lapelinc Framework
	1 Introduction
	2 The Lapelinc Framework
	3 Use of the Lapelinc Framework for the Systematic Construction of Multiple Types of Corpora of Portuguese
	4 Conclusions
	References
Revisiting CCNet for Quality Measurements in Galician
	1 Introduction
	2 Background and Related Work
	3 Methodology
	4 Analysis and Results
		4.1 Error Analysis
		4.2 Heuristics
	5 Conclusion and Future Work
	References
Identifying Literary Characters in Portuguese
	1 Identifying Characters in Literary Text
		1.1 Brief Motivation from Literary Studies
		1.2 Motivation from Computer Science Studies
	2 Previous Work
	3 The DIP Setup
	References
Should I Buy or Should I Pass: E-Commerce Datasets in Portuguese
	1 Introduction
	2 Related Works
	3 Dataset Creation
		3.1 Data Statistics
		3.2 Creating the Dataset
	4 Baseline Results
		4.1 Methodology
		4.2 Results
	5 Final Thoughts
	References
Best MsC/MA and PhD Dissertation
Abstract Meaning Representation Parsing for the Brazilian Portuguese Language
	1 Introduction
	2 The AMR-Annotated Corpus
	3 AMR Parsing
	4 The Evaluation Measure
	5 The Alignment
	6 Final Remarks
	References
Enriching Portuguese Word Embeddings with Visual Information
	1 Introduction
	2 Objectives and Related Work
	3 Methodology
	4 Findings
	5 Impact and Conclusions
	References
Author Index