This dissertation analyzes the impacts of algorithmic discrimination in automated hiring processes, particularly within labor relations, through an interdisciplinary approach that encompasses legal, technical, and ethical aspects. It begins with the recognition that Artificial Intelligence (AI) and Machine Learning (ML) systems are increasingly used in recruitment processes, offering promises of efficiency and impartiality. However, far from being neutral, these systems can reproduce and even intensify historical inequalities, disproportionately affecting vulnerable groups based on gender, race, age, and social class. The methodology involves bibliographic and documental research, with critical analysis of national and international legislation, and specialized literature in the fields of Law, Computer Science, and Ethics. In the legal domain, the study examines the normative framework that guarantees equality and non-discrimination in labor relations, highlighting the constitutional principles of human dignity, substantive equality, and the prohibition of discrimination. It also connects the discussion to the United Nations 2030 Agenda for Sustainable Development, particularly SDGs 5, 8, and 10. From a technical perspective, the study explores the foundations of algorithms used in recruitment, types of machine learning, and the critical role of training data, illustrating how biases can be introduced at every stage of system development. Practical examples of algorithmic discrimination are presented, along with the risks posed by opaque systems ("black boxes") and their social effects on vulnerable populations. The findings indicate an urgent need for specific regulation to ensure the ethical and responsible use of AI in recruitment. The dissertation evaluates strategies for bias mitigation, labor compliance practices, algorithmic auditing, and explainability techniques, showing that technological fairness requires proactive governance aligned with fundamental rights. It concludes that algorithmic discrimination is not an inevitable side effect of technology, but rather a consequence of human decisions that can, and must, be regulated, audited, and corrected. The study proposes pathways to build fairer and more transparent recruitment systems, consistent with constitutional principles and international commitments to equality.

In view of the growing presence of the field of Artificial Intelligence in English language
teaching in basic education, especially through large-scale language models (LLMs), this research
investigates whether systems such as ChatGPT – OpenAI’s GPT-5 model, DeepSeek-V2, and
Gemini 2.5 Flash reproduce prejudiced biases when used in English language teaching for
Brazilian students. Unlike studies that assume the prior existence of such biases, this study
adopts an empirical and exploratory perspective, seeking to verify the existence, or possible
absence, of linguistic, social, and cultural stereotypes in the responses generated by these
generative AI systems. For this purpose, a combined methodology is applied, involving the
Holistic Evaluation of Language Models (HELM), Behavioral Testing with a Checklist, and the
Datamorphic Test. Ten prompts were developed based on real classroom situations, focusing on
sensitive categories such as gender and professions, ethnicity and nationality, social class and
criminality, appearance and body. The responses are analyzed regarding the presence or absence
of stereotypes, representational inequalities, and the omission of cultural plurality. The study
seeks to understand to what extent these technologies can contribute to an inclusive and critical
English language teaching, or, on the contrary, whether they reinforce exclusionary hegemonic
narratives. The results aim to support educators and researchers in ethical and pedagogical
decision-making regarding the use of AI in the classroom, indicating potentialities, limitations,
and recommendations for responsible mediation.

 This dissertation examines the structural disconnect between adherence to open science standards and multimodal accessibility in institutional open science repositories, arguing that this disconnect constitutes a form of exclusion for individuals with low functional vision. The central hypothesis posits that the foundational architectural integration of Digital Humanities, FAIR principles, and multimodality is an epistemological condition—rather than a mere technical adjustment—for the substantive scientific participation of this user group. The methodology combines a systematic literature review with a case study of the ARCA DADOS repository (Fiocruz), simultaneously evaluating the repository's compliance with FAIR principles and WCAG 2.1 guidelines, as well as its multimodal suitability for users with low functional vision. The Umwelt software serves as a reference model, demonstrating that integrating sonification, textual description, and structured editing can form the architectural basis for inclusive scientific ecosystems. The results reveal a significant gap between the openness claimed by ARCA DADOS and the actual cognitive participation it affords users with low vision, replicating the pattern of "declaratory openness without cognitive effectiveness" documented in the literature. The study’s original contribution is an applied diagnostic framework capable of simultaneously assessing FAIR compliance, WCAG accessibility, and multimodal suitability in Brazilian public repositories, while proposing guidelines to bridge the gap between normative openness and cognitive equity.

OLIVEIRA, Fernanda Silva de. Race, vulnerability, and mortality: death profiles of black
people in the Rio de Janeiro metropolitan area. 2025. 51p. Dissertation (Master in Digital
Humanities) Instituto Multidisciplinar, Universidade Federal Rural do Rio de Janeiro, Nova
Iguaçu, RJ, 2025.
This study investigates racial disparities in mortality, focusing on the deaths of Black and
Mixed individuals in the Rio de Janeiro Metropolitan Region, specifically analyzing cases of
maternal death, breast cancer, and death by external causes. The primary objective was to
identify mortality patterns and profiles associated with factors such as race/color, educational
attainment, age group, and marital status, employing a cluster analysis methodology with the
K-Modes algorithm to group the data and explore existing disparities. Secondary mortality
data from the Mortality Information System (Sistema de Informação sobre Mortalidade -
SIM) for the period from 2015 to 2022 were utilized. The results revealed the formation of
distinct profiles, highlighting the vulnerability of Black and Mixed women with lower
educational attainment to maternal and breast cancer mortality, as well as the exposure of
young Black and Mixed individuals to lethal violence, thereby reinforcing the influence of
socioeconomic and racial factors on mortality patterns. It is concluded that racial inequality is
a significant determinant of mortality, necessitating targeted public policies and further studies
to understand and mitigate these disparities.

The rapid spread and adoption of artificial intelligence technologies, especially
language models, represents a milestone in the evolution of contemporary digital society. These models, powered by advanced algorithms and trained on large
volumes of data, have the potential to transform a wide range of industries, from
industry to public services. The scalability and versatility of these solutions have
leading them to exponential use (HAAN, 2023) in various applications, creating a scenario where interaction between humans and machines is increasingly common and intrinsic to everyday life.

The present work aims to evaluate biases in language models, especially
large-scale ones, and their impact on the production of contemporary knowledge. O
study seeks to identify how these models reproduce prejudices and stereotypes
from training data, influencing automated decisions and dissemination of
information. Furthermore, it highlights the need for regulation in the area and the importance
of cultural representation in data sets to promote the production of
more equitable and socially responsible knowledge.

Historically, the iron and steel production chain has always presented risks to the
environment due to the nature of its own operation, whether through high greenhouse gas
emissions, soil, air and water degradation, or the intoxication of the biodiversity in its
surroundings. However, the activities of these production chains can also trigger several social
problems, since their industrial bases are generally located in peripheral regions, far from public
attention and lacking inspection. As a consequence, the challenges related to the fight against
extreme poverty, reduction of hunger, pollution and crime are even more accentuated, making
the conditions of the population living in the surroundings of industries in the sector precarious.
The challenges posed by the iron and steel production chain are largely related to the 2030
Agenda, which seeks, through international cooperation, to reduce social inequality, protect
human rights, preserve biodiversity, conserve the environment, and strengthen global
partnerships to foster sustainable development.
The 2030 Agenda, signed in September 2015 during the United Nations Summit on
Sustainable Development (New York, United States) by 193 Member States, is categorized into
17 Sustainable Development Goals (SDGs), which together add up to 169 targets: (1) No
Poverty, (2) Zero Hunger, (3) Good Health and Well-Being, (4) Quality Education, (5) Gender
Equality, (6) Clean Water and Sanitation, (7) Affordable and Clean Energy, (8) Decent Work
and Economic Growth, (9) Industry, Innovation and Infrastructure, (10) Reduced Inequalities,
(11) Sustainable Cities and Communities, (12) Responsible Consumption and Production, (13)
Climate Action, (14) Life Below Water, (15) Life on Land, (16) Peace, Justice and Strong
Institutions, and (17) Partnerships for the Goals.
The purpose of this work is to understand if companies operating in the iron and steel
production chain are meeting the objectives set out in the 2030 Agenda, or if themes inherent
to sustainable development are being used mainly as an advertising attraction. Do companies
in the industry follow the 17 SDGs or do they have their own agenda? Are all 17 SDGs treated
with their due importance, or is there a greater tendency to prioritize/neglect a specific SDG?
Is there any congruence in the actions practiced by companies in this industrial sector? To
answer these questions, three companies in the same production chain were selected: Vale,
Gerdau and MRS.
The choice of the cut occurred due to the similarities between the companies. Both
operate in the same production chain, carrying out similar and/or codependent industrial

8

activities, and operate in the same geographic space: the extreme west of the metropolitan
region of Rio de Janeiro, the Baixada Fluminense. The Baixada, which is geographically close
to the municipality of Rio de Janeiro, is often forgotten by the government.
To carry out the work, an automated reading of the Annual Report and the Sustainability
Report for the year 2021 of the three chosen companies, Vale, MRS and Gerdau, was carried
out, totaling six documents. The reference year chosen was 2021 because it was the most recent
report at the time the survey had started.
Methodologically, the first step was to understand what the main terms of the 2030
Agenda were and, subsequently, to use them as a guide in company reports through an artificial
intelligence-based approach. The objective was to understand whether the results obtained
through an experiment created based on topical reading were in congruence with the terms
present in the Agenda 2030.
The importance of this approach is mainly due to the lack of supervision in companies
in the face of the commitments established in the United Nations global compact, and due to
the sensationalism often used to overvalue sustainability actions, whose impact is little noticed
by the population living around these industrial activities. Although the experiment is not able
to state whether the actions carried out by Vale, MRS and Gerdau were aligned with the 17
SDGs and their respective 169 targets, it manages to make it possible to make possible which
terms were most used, making it possible to compare these results with the 2030 Agenda, and
between the companies themselves. This also helps us to understand if there is any SDG with
greater or lesser emphasis on the results obtained after automated reading, without human bias.
Through Digital Humanities, it was possible to align a vast bibliographic repertoire and
the critical nature of the Human Sciences with different techniques of data mining, collection
and analysis, enabling technological interventions, natural to Computer Sciences. Although
there is still room for further adjustments and improvements, the multidisciplinarity fostered by
Digital Humanities has made it possible for documents to be analyzed in an impartial manner,
being a kick for new research that uses large data frameworks and/or that requires neutrality
from the researcher.

The internet has brought, among other advances, agility, and ease of use of distributed
services, the massive production of data has changed and continues to transform
relationships between people, organizations, and governments. However, despite of these
benefits, criminal organizations have also evolved and adapted, starting to operate
intensively in this new digital territory, including taking advantage of the global reach of
their actions, geopolitical instabilities, and the difficulties of national States in protecting
their citizens against cybercrimes. This dissertation has a transdisciplinary approach, it
seeks to investigate the intersection of the Digital Humanities and the process of digital
transformation, focusing on the growing phenomenon of cybercrime. In this work we
developed a method to create a dataset related to cybercrimes that occurred between the
years 2000 and 2021, we collected real data from more than 30 Brazilian public security
institutions. Data were analysed qualitatively and quantitatively to understand the
Brazilian scenario of cybercrime. Additionally, this dissertation also presents and
discusses a conceptual proposal of a computational artifact that can be considered as an
initial contribution to increase cyber resilience of Brazilian State or augment cyber
security of public institutions to support citizens regarding the mitigation of problems
related to most serious cybercrimes.

The work deals with how Brazilian law has adapted to the inclusion of artificial intelligence within the penitentiary system. The work deals especially with the relationships between technology developments and the repercussions of ADPF 347.

In the context of the Big Data era, where resources are increasingly dataified and processes are digitized, there is a growing challenge in ensuring the quality, reliability, and security of data. Human efforts alone are insufficient to address these concerns. Consequently, various opportunities have arisen, leveraging processes, principles, and emerging technologies to offer more suitable data treatments and effective ways of managing and sharing scientific data. Two prominent approaches in this context are the Semantic Web and the more recent FAIR Data Principles. These frameworks not only advocate for the utilization and reusability of data and metadata but also align with the Linked Open Data movement and its underlying principles. In essence, they facilitate the dissemination of data interconnection practices and promote their availability on the data web. Presently, the data repository of the OpenSoils platform contains a substantial volume of pedological data with significant relevance across multiple knowledge domains, including digital agriculture, sustainability, and digital humanities. However, this repository is not yet fully integrated into the data web. This research explores the intersections of concepts related to the Semantic Web, FAIR Principles, and Linked Open Data movements within the digital humanities domain. It proposes the development of intelligent artifacts utilizing ETLH (Extract, Transform, Load, Harmonize) workflows and intelligent algorithms. These artifacts are envisioned to be applied to the OpenSoils platform data repository in the context of digital humanities. The overarching goal is to establish a new digital data repository for the OpenSoils platform, characterized by its transdisciplinary nature and based on RDF (Resource Description Framework) triples. The origin of this repository is rooted in a project within the Digital Agriculture domain.

Technological advances in recent decades have allowed digital information, available in the virtual environment, to become an efficient means of transmitting recorded knowledge. In this context, digital repositories (RD) emerge as a virtual database that stores, preserves, organizes and provides access to the electronic collection, also playing an important role in providing visibility to the documentary production of institutions. In 2005, the Federal Rural University of Rio de Janeiro (UFRRJ) joined the Brazilian Digital Library of Theses and Dissertations (BDTD), an exclusive digital repository of these documents. With the growth in the volume of these academic works at UFRRJ, a worrying question arises: how to efficiently manage this electronic collection of the institution in its BDTD repository, with the objective of increasing the visibility of its content and, consequently, promoting greater use by the academic and scientific community? The general objective of this study was to present proposals to improve the treatment of digital theses and dissertations at UFRRJ, seeking solutions to several issues related to the processing of items in the BDTD. The research developed a qualitative empirical approach, divided into two phases. The first stage consisted of a review of the literature relevant to the research problem, with emphasis on the topics: management of digital repositories, processing of electronic theses and dissertations, organization of information, Ranganathan's Laws, FAIR Principles and the study of the relationship between Digital Humanities and Information Science. In the second phase, based on the knowledge acquired from the selected sources, a detailed analysis was carried out in the BDTD, with the aim of applying theories, procedures and techniques that allow the systematization of the informational treatment of theses and dissertations in this information system. Thus, an action-research approach was used, encompassing observation and active intervention in the object of study. It was concluded that this virtual environment needs a broader and interdisciplinary approach to guarantee a better experience for users in their search for knowledge.

Schools still face difficulties in promoting adequate stimuli for the development of unique skills and talents among individuals with intellectual disabilities. The limited use of information and communication technologies in the daily lives and in the teaching-learning process of these young adults is a limiting aspect for their integration into the job market. This becomes even more challenging considering the social vulnerability and deficiencies experienced by these individuals and their families. Often, the sole family income comes from the Continuous Cash Benefit (BPC) for the person with intellectual disabilities, leading them to exceed their caregiving responsibilities, thus depriving these individuals of their autonomy. Society imposes barriers through prejudice, discrimination, and a lack of information about disabilities, leaving them on the sidelines of social relationships and denying their collaborative abilities. These paradigms persist within the collective consciousness, hindering inclusion and rendering people with intellectual disabilities invisible. The research aimed to understand the perspectives of students with intellectual disabilities regarding their integration into the job market through semi-structured interviews conducted at a school in Baixada Fluminense. Through the analysis of the interviewees' responses, it was evident that the job market is perceived as distant from their reality. There is limited use of information and communication technologies, which could serve as tools for developing autonomy, socio-emotional skills, and important technical abilities crucial for students' transition into the job market.

The year 2020 will be a milestone in world history due to the illness caused by the SARS-CoV-2 virus, Covid-19. Isolation policies have been implemented to put a brake on the spread of the disease. Consequently, the brazilian education, under the guidance of ordinance nº 343 of March 17, 2020, substituted face-to-face classes by online classes carried out on digital platforms. Therefore, this research addresses the issues of inclusive education from the perspective of Digital Humanities, checking the promotion of accessibility of technological resources used by deaf and hearing-impaired students during the period of online classes at the Federal Rural University of Rio de Janeiro. The term Digital Humanities has been gaining visibility in the academic ambience for being an interdisciplinary research field. The choice of qualitative approach is related to the investigation carried out in real context on the social sphere. Through the case study method, we were able to gather data from a semi-structured questionnaire in which the questions were asked and answered in an interview granted to us by the subjects of this research. With the aid of the ATLAS.ti software, which uses the analytical methods of Bardin (2011), we analyzed the data from two resources made available in this program, networks and words cloud. After systematizing the data, we evidenced that the participation of deaf students in online classes was possible, however, imbued with many difficulties regarding accessibility and that the presence of the Libras interpreter is not a guarantee to make this issue effective. For students with hearing impairment, accessibility was only possible with the help of assistive technologies associated with digital platforms.

Publications in scientific journals and in specialized conferences play a key role in expressing the topics of interest to authors and readers in a given field of knowledge. In this sense, the effort to organize scientific production is vital for the advancement of the dissemination of contents produced in an unequivocal, fast and safe way. Considering the current information flood caused by digital tools, the issue of automated classification becomes urgent and must be addressed in every repository or digital platform of scientific publications. Among other aspects, the use of a taxonomy stands out for its ability to add a hierarchical semantic element to the act of classifying or categorizing concepts and specific information that define the domain of a field of knowledge. Particularly in the field of Digital Humanities, the epistemological culture that has been built by its growing community has given rise to international projects that address the issue in an environment with additional challenges due to its strongly interdisciplinary profile. The objective of this dissertation is to use computational tools for analysis by topics of texts to develop an auxiliary method of lexical classification of publications supported by a taxonomy called {\em TaDiRAH -- Taxonomy of Digital Research Activities in the Humanities}. The proposed method can be seen as a combination of the semantic approach of taxonomy with the lexical approach of automated text analysis. Its categories are free and practical. However, it is not uncommon, and even expected by the interdisciplinary profile, that a publication can be classified into different categories of different levels or the same level of the taxonomy, thus creating overlaps. In addition, the number of publications already classified by the scientific community is still relatively small and, above all, extremely unbalanced between the taxonomy categories. These two aspects that characterize the available sample make the task of reliably classifying publications in Digital Humanities particularly difficult. We propose a method that combines Bayesian classification models from the literature with original approaches to deal with overlaps and imbalances between taxonomy categories. Results of computational experiments carried out with a universe of 443 publications showed that the proposed approaches are, in fact, capable of profoundly improving the performance of the classification methods use.

The idea of sharing hosting services having as its main intermediary the Airbnb digital platform has a great discursive appeal about the possibility of providing tourists with a more personalized, authentic and genuine experience in the daily life of the cities visited. The so-called sharing economy has taken advantage with relative success of the potential for global interaction in the digital world made possible by the spread of the internet, to propose a business model considered to be totally disruptive. Thinking about the theoretical-methodological paths that the relationship between Digital Humanities and Geography make possible for the conformation of a critical thematic analysis, we proposed to assess whether, in the case of the city of Rio de Janeiro, the main port of entry for tourists in Brazil and the country's largest leisure tourist destination, the logic of sharing residential units for tourism purposes, in fact, enhances a disruptive spatial rebalancing in the provision of services for tourism in this city, historically characterized by alarming rates of economic concentration in its South Zone. The development of the dissertation is based on exploratory qualitative-quantitative research procedurally structured from a multidisciplinary bibliographic and documentary review mixed with critical analysis and interpretation of data, aiming to provide an overview of the issues related to the theme, deepening the discussion on the profile of the phenomenon in the area of and study. We hope, at the end of the research, to have worked with the real dimension of the phenomenon of economic, spatial and legal disruption that the sharing economy can cause wherever companies in the sector are installed; we also hope to be successful in a theoretically in-depth discussion on the nature of Digital Humanities as a field of knowledge and an area of scientific knowledge, in addition to evaluating how the relationship between technological development and knowledge production in the humanities can develop under a non-exclusive logic.

In times of institutional crisis between the Executive and Judiciary powers in Brazil,
the call for research that analyzes the behavior of ministers of
Federal Court of Justice. Studies on judicial behavior involve theories and
statistical data in search of concrete answers about decision-making patterns
individual and collective courts. Topics such as spatial voting models,
theories of judicial behavior and ideal point estimation are widely
addressed in international research and are beginning to permeate the Brazilian scenario.
As recent works related to the highest instance of the judiciary
Brazilians investigate processes related to the concentrated control of
constitutionality, the present research intends to present a new approach: the
decision patterns implicit in the vote scores related to requests for
Habeas Corpus judged by the Federal Supreme Court from 1988 to 2021.
The objective is to identify ideological/legal patterns and groupings among the
analyzed voters, having as a source of data all the Habeas processes
Corpus voted by the Plenary of the Brazilian Supreme Court in the analyzed period. With
investigations into the role of the court of last resort, the research contributes with
the production of a collection of data on decisions and aims to present the
inferences from the analysis of this dataset, with answers to some
questions and debates inherent to studies on supreme courts, assuming
characteristics of a typical Digital Humanities work. are investigated the

possible coalition formations, presence of outliers, political-political influences
parties, in addition to guarantor traits in collective and individual votes. At

investigations are divided into seven time frames and, in this way, the
answers on how the geography of the Federal Supreme Court was presented in
each period examined.

The objective of this research was to analyze how the Capes and Ibict databases address the issue of the congenital zika virus syndrome (SCZV) and its dissemination considering the Digital Humanities (HDs). HDs have been operating today in the production of knowledge, as well as political, ideological and, above all, intellectual resignification in terms of scientific production and dissemination using digital technology methodologies. It is a relatively new interdisciplinary area, little studied in our country. In this sense, this research is relevant and justified by addressing Special Education from the perspective of Digital Humanities. It is a qualitative research, of a bibliographic type. This investigation is linked to the research project "Research and intersectoral actions between education and health in promoting schooling and the development of children with congenital Zika virus syndrome in the Baixada Fluminense" funded by the Research Support Foundation of the State of Rio de Janeiro ( FAPERJ). For data analysis we used the analysis by Bardin (2011) through the categorization process and with the application of the ATLAS ti software. The data, among other aspects, show that the majority (52.17%) of the surveys were carried out in the southeastern region of the country at universities and public research institutions. They also showed that most of them are from the area, but that they also had studies in other fields, such as social assistance, but only one focuses on more specific aspects of the educational area. It was also clear the importance of digital and computational tools for carrying out the analyzed studies, as well as free access to databases on the subject.

In the search for academic articles based on the Theory of Moral Foundations applied to
texts in Portuguese, the scarcity of scientific material on the subject is noted.
The analysis proposed in this research deals with the identification of some characteristics.
cultural stics of Portuguese-speakers, mainly from the point of view of
morality of these individuals. It also discusses the opinions of the collective of active individuals
on the social network Twitter, and highlights the characteristics of discursive narratives in this environment. It is known that the language used by individuals is a reflection of their thoughts, emotions,
values, culture, and a means of communication and interaction with other members of society.
Therefore, analyzing the content of digital texts from Brazilian Portuguese speakers is necessary
to identify the expressions of moral values present in the texts. For this, the
exploratory research method, with the technical procedure of analysis of ethical content
in association with topic modeling as a natural language processing technique.
We collected 230,215 textual data from interactions of users of the social network Twitter, for the period
April 14-20, 2021, containing at least one of the hashtags: #caemcasa, #pandemic,
#forabolsonaro, #closedcombolsonaro. Textual expressions associated with values were analyzed.
morals of the subjects who interacted in this social network from the tweets collected, al in
of the general characteristics of these texts. Finally, it was concluded that textual data from networks
Virtual social media are rich content for observing the behavior and personal opinions of
Portuguese-speaking individuals on controversial issues such as isolation
social in times of pandemic and as a political position.