Brazilian Portuguese Texts Research Articles

CachacaNER: a dataset for named entity recognition in texts about the cachaça beverage

Named Entity Recognition (NER) is the task of identifying and classifying tokens in texts corresponding to a set of pre-defined categories, such as names of people, organizations and locations. Datasets labeled for this task are essential for training supervised machine learning models. Although there are many datasets labeled with texts for English, in the Portuguese language they are scarcer. This work contributes to the creation and evaluation of a manually labeled dataset for the NER task, with texts in Brazilian Portuguese, in the specific domain of the beverage called Cachaça. This is a popular drink in Brazil, and of great economic importance. This is the first NER dataset in the beverage domain, and can be useful for other types of beverages with similar entity categories, such as wine and beer. We describe the process of data collection, creation of the dataset and its experimental evaluation. As a result, we created a dataset containing over 180,000 tokens labeled in 17 entity categories. The labeling obtained an agreement coefficient of 0.857 among the labelers, according to the Fleiss’ Kappa metric, which is considered almost perfect. In our experimental evaluation, we obtained a micro-F1 value equal to 0.933 in the test set. The size of the dataset, as well as the result of its experimental evaluation, are comparable to other datasets in the Portuguese language, even though ours has a greater number of entity categories.

Words Similarities on Personalities: A Language-Based Generalization Approach for Personality Factors Recognition

The evaluation of personality traits allows the study of human behavior in different environments, but it is not a trivial task. In this sense, the Five-Factor Model (FFM) allows, in a global way, the assessment of personality traits of individuals using textual data. However, there is a scarcity of lexical resources for languages other than English, which generated the main research question of this work: "Can models trained to predict FFM personality traits using English textual data show satisfactory results when applied to textual data in other languages?". Therefore, this work aims to answer: (i) Whether Word Embeddings techniques could be used to solve low resources languages problems in FFM personality traits prediction; and (ii) Whether is feasible to train a traditional Machine Learning algorithm with English language textual data and evaluate its performance with Brazilian Portuguese language textual data for FFM personality traits prediction. Thus, the work aims to present an approach in which the models can be used to learn the highest level of abstraction. As results, we observed that the difference in performance between the models trained for personality recognition in English is minimal when used to predict FFM personality traits in Brazilian Portuguese texts. In this task, the Stochastic Gradient Descent model presents the best average results among the FFM personality traits of the models analyzed.

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

Text classification is a natural language processing (NLP) task relevant to many commercial applications, like e-commerce and customer service. Naturally, classifying such excerpts accurately often represents a challenge, due to intrinsic language aspects, like irony and nuance. To accomplish this task, one must provide a robust numerical representation for documents, a process known as embedding. Embedding represents a key NLP field nowadays, having faced a significant advance in the last decade, especially after the introduction of the word-to-vector concept and the popularization of Deep Learning models for solving NLP tasks, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and Transformer-based Language Models (TLMs). Despite the impressive achievements in this field, the literature coverage regarding generating embeddings for Brazilian Portuguese texts is scarce, especially when considering commercial user reviews. Therefore, this work aims to provide a comprehensive experimental study of embedding approaches targeting a binary sentiment classification of user reviews in Brazilian Portuguese. This study includes from classical (Bag-of-Words) to state-of-the-art (Transformer-based) NLP models. The methods are evaluated with five open-source databases with pre-defined data partitions made available in an open digital repository to encourage reproducibility. The Fine-tuned TLMs achieved the best results for all cases, being followed by the Feature-based TLM, LSTM, and CNN, with alternate ranks, depending on the database under analysis.

Evaluating a typology of signals for automatic detection of complementarity

In a cluster of news texts on the same event, two sentences from different documents might express different multi-document phenomena (redundancy, complementarity, and contradiction). Cross-Document Structure Theory (CST) provides labels to explicitly represent these phenomena. The automatic identification of the multi-document phenomena and their correspondent CST relations is definitely handy for Automatic Multi-Document Summarization since it helps computers understand text meaning. In this paper, we evaluated a typology of (textual) signals for the automatic detection of the CST relations of complementarity (i.e., Historical background, Follow-up and Elaboration) in a multi-document corpus of news texts in Brazilian Portuguese. Using algorithms from different machine-learning paradigms, we obtained classifiers that achieved high general accuracy (higher than 90%), indicating the potential of the signals.

RastrOS Project: Natural Language Processing contributions to the development of an eye-tracking corpus with predictability norms for Brazilian Portuguese.

This article presents RastrOS, a new eye-tracking corpus of eye movement data from university students during silent reading of paragraphs of texts in Brazilian Portuguese (BP). The article shows the potential of the corpus for natural language processing (NLP) using it to evaluate the sentence complexity prediction task in BP and it also focuses on the description of NLP resources and methods developed to create the corpus. Specifically, we present: (i) the method used to select the corpus paragraphs from large corpora, using linguistic metrics and clustering algorithms; (ii) the platform for collecting the Cloze test, which is also responsible for creating the project datasets, and (iii) the hybrid semantic similarity method, based on word embedding models and contextualised word representations, used to generate semantic predictability norms. RastrOS can be downloaded from the open science framework repository with the computational infrastructure mentioned above. Datasets with predictability norms of 393 participants and eye-tracking data of 37 participants are available in the OSF repository for this work (https://osf.io/9jxg3/).

Multilingual Corpus on Migration and Asylum (COMMIRE)

In recent years, the world has faced an unprecedented migration crisis. "Mobilidades e contatos de línguas" (MOBILANG), a research group at the University of Brasilia focusing on the study of language in the context of migration and asylum, has identified that language is the most significant barrier migrants face in accessing essential services after arriving in Brazil. To mitigate this problem, the group envisioned the creation of a multilingual corpus-driven glossary on the subject aimed at translators, interpreters, those who work with migration, and migrants. The glossary is a partnership with the research group "Terminologia e Tradução Direcionadas por Corpus" (TermiTraDiCo), and has the support of the UNESCO Chair on Language Policies for Multilingualism. The initial stage consisted of compiling a robust electronic corpus to serve as the base for the glossary. This paper aims to describe the planning, compilation, and overall content of this corpus, named Corpus Multilíngue sobre Migração e Refúgio (Multilingual Corpus on Migration and Asylum) – COMMIRE, a medium-large corpus containing comparable (and some parallel) texts in Brazilian Portuguese, Spanish, French, and English.

Information Retrieval Based on Brazilian Portuguese Texts

Knowledge-based intelligent systems might be used in the banking sector to automate customer service. One of the ways to represent knowledge that is both understandable by humans and readable by machines is by using ontologies. Whenever a customer queries its bank regarding specific products or services, the existing knowledge modeled in an ontology might be used by a customer service chatbot to answer it in an automated way. The existing manual information retrieval process from banking specialists is laborious and time-consuming. Specialists use natural language, visual representations, and common sense, often overlooking details. It is a great challenge to make a specialist's knowledge explicit, formal, precise, and completely scalable, which is the format required by a customer service chatbot. We propose a semi-automatic approach to retrieving banking information in Brazilian Portuguese texts with minimal specialist support. By combining Natural Language Processing techniques (e.g., syntactic analysis to obtain the logical meaning of sentences based on rules and its structure) and an ontology constructor library, it was possible to build a tool that receives texts from the banking domain and constructs an ontology that knowledge-based intelligent systems can use. Specialist support is only needed in intermediate refinement steps, thus optimizing the banking specialist's time. The use cases for investments, opening a banking account, and the comparison of the proposed approach show how we reduced manual labor in the information retrieval process by a factor of 40%. Our approach can identify more information in each sentence compared to a similar method found in the literature. The resulting ontologies can be used in a chatbot that automates customer support for a large Brazilian bank.

Performing Multilingual Analysis With Linguistic Inquiry and Word Count 2015 (LIWC2015). An Equivalence Study of Four Languages.

Today, there is a range of computer-aided techniques to convert text into data. However, they convey not only strengths but also vulnerabilities compared to traditional content analysis. One of the challenges that have gained increasing attention is performing automatic language analysis to make sound inferences in a multilingual assessment setting. The current study is the first to test the equivalence of multiple versions of one of the most appealing and widely used lexicon-based tools worldwide, Linguistic Inquiry and Word Count 2015 (LIWC2015). For this purpose, we employed supervised learning in a classification problem and computed Pearson's correlations and intraclass correlation coefficients on a large corpus of parallel texts in English, Dutch, Brazilian Portuguese, and Romanian. Our findings suggested that LIWC2015 is a valuable tool for multilingual analysis, but within-language standardization is needed when the aim is to analyze texts sourced from different languages.

Defining a state-of-the-art POS-tagging environment for Brazilian Portuguese clinical texts

Natural language processing techniques are essential for unlocking patients’ data from electronic health records. An important NLP task is the ability to recognize morphosyntactic information from the texts, a process called part-of-speech (POS) tagging. Currently, neural network architectures are the state-of-the-art method, although there is a lack of studies exploiting this approach within Brazilian Portuguese clinical texts. The objective of this study is to define a state-of-the-art POS-tagging environment for Brazilian Portuguese clinical texts. We reviewed multiple neural network-based POS-tagging algorithms, and the Flair tool was selected due to its exceptional performance in the journalistic domain, as there is any specific algorithm to Portuguese clinical texts. We executed a normalization process on available corpora from multiple domains (two journalistic, one biomedical, one clinical, and a new corpus composed of all three of these). The Flair algorithm was trained with all corpora, generating five models, which were evaluated with all domains. The clinical model achieved 92.39% accuracy (previous POS-tagging clinical work reached 91.5%); the biomedical model achieved 97.9% accuracy. All the models were assessed on their own test set. We developed a new state-of-the-art modeling environment for POS tagging of Brazilian Portuguese clinical texts and achieved comparable results to other state-of-the-art studies in journalistic contexts.

Detecting the Central Units of Brazilian Portuguese argumentative answer texts

Understanding or writing properly the main idea or the Central Unit (CU) of a text is a very important task in exams. So, detecting automatically the CU may be of interest in language evaluation tasks. This paper presents a CU detector based on machine learning techniques for argumentative answer texts in Brazilian Portuguese. Results show that the detection of CUs following machine learning techniques in argumentative answer texts is better that those using rules.

A text summarization method based on fuzzy rules and applicable to automated assessment

In the last two decades, the text summarization task has gained much importance because of the large amount of online data, and its potential to extract useful information and knowledge in a way that could be easily handled by humans and used for a myriad of purposes, including expert systems for text assessment. This paper presents an automatic process for text assessment that relies on fuzzy rules on a variety of extracted features to find the most important information in the assessed texts. The automatically produced summaries of these texts are compared with reference summaries created by domain experts. Differently from other proposals in the literature, our method summarizes text by investigating correlated features to reduce dimensionality, and consequently the number of fuzzy rules used for text summarization. Thus, the proposed approach for text summarization with a relatively small number of fuzzy rules can benefit development and use of future expert systems able to automatically assess writing. The proposed summarization method has been trained and tested in experiments using a dataset of Brazilian Portuguese texts provided by students in response to tasks assigned to them in a Virtual Learning Environment (VLE). The proposed approach was compared with other methods including a naive baseline, Score, Model and Sentence, using ROUGE measures. The results show that the proposal provides better f-measure (with 95% CI) than aforementioned methods.

Subtopic annotation and automatic segmentation for news texts in Brazilian Portuguese

Subtopic segmentation aims to break documents into subtopical text passages, which develop a main topic in a text. Being capable of automatically detecting subtopics is very useful for several Natural Language Processing applications. For instance, in automatic summarisation, having the subtopics at hand enables the production of summaries with good subtopic coverage. Given the usefulness of subtopic segmentation, it is common to assemble a reference-annotated corpus that supports the study of the envisioned phenomena and the development and evaluation of systems. In this paper, we describe the subtopic annotation process in a corpus of news texts written in Brazilian Portuguese, following a systematic annotation process and answering the main research questions when performing corpus annotation. Based on this corpus, we propose novel methods for subtopic segmentation following patterns of discourse organisation, specifically using Rhetorical Structure Theory. We show that discourse structures mirror the subtopic changes in news texts. An important outcome of this work is the freely available annotated corpus, which, to the best of our knowledge, is the only one for Portuguese. We demonstrate that some discourse knowledge may significantly help to find boundaries automatically in a text. In particular, the relation type and the level of the tree structure are important features.

O DESENVOLVIMENTO DA EXPRESSÃO MODALIZADORA “PODE SER”. UM CASO DE (INTER)SUBJETIVIZAÇÃO NO PORTUGUÊS

Comprometido com um enfoque funcionalista, o trabalho analisa a expressão modalizadora pode ser como predicado encaixador de proposição (pode ser1) e como construção independente (pode ser2), em textos contemporâneos do português brasileiro, de fala e de escrita. Busca-se identificar graus de (inter)subjetividade reveladores de um processo de (inter)subjetivização, conforme proposta de Traugott (2010). Sustentam a análise parâmetros de (inter)subjetividade de elementos modalizadores indicados especialmente em Traugott e Dasher (2002) e a noção de modalidade como categoria multifuncional, que não apenas codifica atitude do falante em relação ao conteúdo modalizado, mas que também atua como estratégia pragmática, como reguladora da situação comunicativa. A pesquisa revela pode ser como uma fórmula de grande aproveitamento no jogo discursivo, um conjunto bastante solicitado, produtivo e útil nas relações interpessoais. O estudo de propriedades semânticas, discursivas e morfossintáticas indica deslizamento na operação da construção, da sintaxe (pode ser1) para o discurso (pode ser2), interpretado como próprio de (inter)subjetivização.

Spelling Error Patterns in Brazilian Portuguese

Fifty years after Damerau set up his statistics for the distribution of errors in typed texts, his findings are still used in a range of different languages. Because these statistics were derived from texts in English, the question of whether they actually apply to other languages has been raised. We address this issue through the analysis of a set of typed texts in Brazilian Portuguese, deriving statistics tailored to this language. Results show that diacritical marks play a major role, as indicated by the frequency of mistakes involving them, thereby rendering Damerau's original findings mostly unfit for spelling correction systems, although still holding them useful, should one set aside such marks. Furthermore, a comparison between these results and those published for Spanish show no statistically significant differences between both languages—an indication that the distribution of spelling errors depends on the adopted character set rather than the language itself.

Automated Text Clustering of Newspaper and Scientific Texts in Brazilian Portuguese: Analysis and Comparison of Methods

This article reports the findings of an empirical study about Automated Text Clustering applied to scientific articles and newspaper texts in Brazilian Portuguese, the objective was to find the most effective computational method able to cluster the input of texts in their original groups. The study covered four experiments, each experiment had four procedures: 1. Corpus Selections (a set of texts is selected for clustering), 2. Word Class Selections (Nouns, Verbs and Adjectives are chosen from each text by using specific algorithms), 3. Filtering Algorithms (a set of terms is selected from the results of the preview stage, a semantic weight is also inserted for each term and an index is generated for each text), 4. Clustering Algorithms (the clustering algorithms Simple K-Means, sIB and EM are applied to the indexes). After those procedures, clustering correctness and clustering time statistical results were collected. The sIB clustering algorithm is the best choice for both scientific and newspaper corpus, under the condition that the sIB clustering algorithm asks for the number of clusters as input before running (for the newspaper corpus, 68.9% correctness in 1 minute and for the scientific corpus, 77.8% correctness in 1 minute). The EM clustering algorithm additionally guesses the number of clusters without user intervention, but its best case is less than 53% correctness. Considering the experiments carried out, the results of human text classification and automated clustering are distant; it was also observed that the clustering correctness results vary according to the number of input texts and their topics.

Extractive summarization using complex networks and syntactic dependency

The realization that statistical physics methods can be applied to analyze written texts represented as complex networks has led to several developments in natural language processing, including automatic summarization and evaluation of machine translation. Most importantly, so far only a few metrics of complex networks have been used and therefore there is ample opportunity to enhance the statistics-based methods as new measures of network topology and dynamics are created. In this paper, we employ for the first time the metrics betweenness, vulnerability and diversity to analyze written texts in Brazilian Portuguese. Using strategies based on diversity metrics, a better performance in automatic summarization is achieved in comparison to previous work employing complex networks. With an optimized method the Rouge score (an automatic evaluation method used in summarization) was 0.5089, which is the best value ever achieved for an extractive summarizer with statistical methods based on complex networks for Brazilian Portuguese. Furthermore, the diversity metric can detect keywords with high precision, which is why we believe it is suitable to produce good summaries. It is also shown that incorporating linguistic knowledge through a syntactic parser does enhance the performance of the automatic summarizers, as expected, but the increase in the Rouge score is only minor. These results reinforce the suitability of complex network methods for improving automatic summarizers in particular, and treating text in general.

A complex network approach to text summarization

Automatic summarization of texts is now crucial for several information retrieval tasks owing to the huge amount of information available in digital media, which has increased the demand for simple, language-independent extractive summarization strategies. In this paper, we employ concepts and metrics of complex networks to select sentences for an extractive summary. The graph or network representing one piece of text consists of nodes corresponding to sentences, while edges connect sentences that share common meaningful nouns. Because various metrics could be used, we developed a set of 14 summarizers, generically referred to as CN-Summ, employing network concepts such as node degree, length of shortest paths, d-rings and k-cores. An additional summarizer was created which selects the highest ranked sentences in the 14 systems, as in a voting system. When applied to a corpus of Brazilian Portuguese texts, some CN-Summ versions performed better than summarizers that do not employ deep linguistic knowledge, with results comparable to state-of-the-art summarizers based on expensive linguistic resources. The use of complex networks to represent texts appears therefore as suitable for automatic summarization, consistent with the belief that the metrics of such networks may capture important text features.

Textos padronizados em português (BR) para medida da velocidade de leitura: comparação com quatro idiomas europeus

To develop standardized texts in Brazilian Portuguese to assess reading speed and compare performances among four European languages. 10 texts were designed by a linguistic expert at the level of a sixth grade reading material (reading ages 10-12 years) and were matched for length and syntactic complexity, according to the syntactic prediction locality theory of Gibson. Normally sighted native speaking volunteers aged 18-35 years read each text aloud at random. The reading speed was on average 1100 +/- 167 (standard deviation) characters per minute. Only small differences were found between the measured reading speeds of the 10 texts, and these differences were not statistically significant in groups of at least 6 texts. The mean reading speed in age-matched volunteers in Finnish was: 1263 +/- 142; French: 1214 +/- 152; English: 1234 +/- 147; and German: 1126 +/- 105. The authors developed a set of standardized, homogeneous, and comparable texts in Brazilian Portuguese. These texts will be a valuable tool to measure reading speed for repeated measurements and in international studies in the field of reading and low vision research.

The research article and the science popularization article: a probabilistic functional grammar perspective on direct discourse representation

This article discusses the results of an investigation on discourse representation in a corpus of 34 million words constituted by texts in Brazilian Portuguese from two different genres: the research article and the science popularization article. Drawing on a systemic functional grammar perspective of language and pursuing a probabilistic approach, it focuses on the realization of lexicogrammatical systems of direct discourse representation as enacting interpersonal and social relationships. It is argued that the citation practices employed by writers in the genres studied negotiate discursive distance and are influenced by the status culturally and historically ascribed to scientific knowledge.

Traduzindo as diferenças extra-lingüísticas – procedimentos e condicionantes

Os marcadores culturais que se manifestam sob forma lingüística nos textos constituem um dos bem conhecidos obstáculos à tradução. São, no entanto, e a despeito das evidentes dificuldades, traduzidos de um modo ou outro. Este trabalho, com base em um estudo conduzido sobre dois textos em português brasileiro e suas respectivas traduções para o inglês norte-americano, investiga os procedimentos tradutórios adotados para os marcadores culturais lexicais, a distribuição dos diversos procedimentos empregados e indaga acerca de possíveis fatores condicionantes de tal distribuição. Os resultados, ao mesmo tempo que confirmam a previsão estrutural, indicam a clara influência das condições de produção sobre a distribuição efetiva das ocorrências.