Privacy regulations impose on companies limitations about the collection, use, and disclosure of user data. One of the actions most companies undertake for this, consists in modifying their systems with processes for consent acquisition and management. Unfortunately, where systems are large and with many dependencies, they often also have little documentation, and knowledge on the system is distributed among different domain experts. These circumstances make the re-engineering of systems a tedious and complex, if not impossible, activity. This PhD Thesis proposes a model-based method with a top-down approach, for modeling consent requirements and analyzing compliance with regulations, by refinement of models from organizational structure down to business processes. The method is provided with guidelines in the form of a process and includes modeling languages and reasoning frameworks for the analysis of requirements with respect to a preset of privacy principles on consent. The Thesis includes validations with realistic scenarios and with domain practitioners from the healthcare domain.

Studien syftar till att undersöka hur tidigare våldsutsatta kvinnor med skyddade personuppgifterupplever sitt medborgarskap och om det finns någon kongruens mellan idén om medborgarskap och kvinnornas egna upplevelser. Utifrån medborgarskapsteori har två dimensioner identifierats: deltagande och känslor. Materialet utgörs av sekundärdata baserad påJämställdmyndighetens rapport 86 gömda kvinnor och deras 128 barn samt berättelser publicerade på organisationen Gömda kvinnors Instagramkonto @gomda.kvinnor. Studiedesignen utgörs av en tvärsnittsstudie där fler än ett fall undersöks över tid för att avslöja mönster i verkligheten. Materialet analyseras med hjälp av en kvalitativ textanalys, mer specifikt en kritisk idéanalys. Resultaten visar att kvinnorna ställs inför många problematiska situationer till följd av skyddet och upplever att samtliga medborgerliga dimensioner i någon mån påverkas av att leva med skyddade personuppgifter. Kvinnornas upplevelser tyder därmed på att det inte finns kongruens mellan idén om medborgarskap och deras egen upplevelse.Inkongruensen tolkas däremot vara störst i en i av subdimensionerna.

[pt] Em um mundo mediado por algoritmos, em que espaços de tomada de decisão antes destinados a humanos passam a ser dominados por estes artefatos, surge uma demanda para que estas decisões algorítmicas sejam explicáveis. Este desafio ganha uma camada de complexidade quando há o uso de técnicas de inteligência artificial, em especial, a aplicação de modelos de aprendizado de máquina, diante da opacidade e inescrutabilidade do modo de funcionamento e dos resultados gerados de alguns tipos destes algoritmos. Neste sentido, esta tese tem início com a apresentação do conceito e dos desafios da inteligência artificial e do aprendizado de máquina para o Direito, particularmente para direitos fundamentais (i.e. proteção de dados pessoais, privacidade, liberdade, autonomia e igualdade). Em seguida, é compartilhada a discussão envolvendo o direito à explicação quando do seu surgimento, e como a sua previsão na LGPD poderá ser interpretada à luz dos aprendizados e interpretações já colhidos no âmbito do GDPR. Ainda, serão analisados como os principais desafios para os direitos fundamentais que são colocados por tais algoritmos de tomada de decisão podem ser resumidos sob os princípios de transparência, prestação de contas e responsabilização e justiça/igualdade. É proposta uma abordagem multifacetada e multidisciplinar, a ser aplicada em diferentes momentos, para assegurar a observância de tais princípios no desenvolvimento e uso de algoritmos de tomada de decisão de aprendizado de máquina. Por fim, propõe-se que a garantia de um direito à explicação, atualmente inserido em uma discussão mais ampla de prestação de contas e responsabilização, deve atender a uma perspectiva de mérito e de procedimento. São identificados os diferentes tipos de conteúdos que têm sido mapeados como passíveis de serem exigidos a título de explicação, e os valores e direitos que um direito à explicação visa proteger, demonstrado, ao final, a importância de que este conteúdo possa estar sujeito a algum tipo de escrutínio público. / [en] In a world mediated by algorithms, in which decision-making spaces previously destined for humans are now dominated by these artifacts, urges a demand for these algorithmic decisions to be explainable. This challenge gains a layer of complexity when artificial intelligence techniques are used, in particular, the application of machine learning models, given the opacity and inscrutability of the operating mode and the results generated by some types of these algorithms. In this sense, this thesis begins with the presentation of the concept and challenges of artificial intelligence and machine learning for the area of Law, particularly for fundamental rights (i.e. data protection, privacy, freedom, autonomy and equality). Then, the discussion involving the arise of a right to explanation is presented, and how its provision in the LGPD can be interpreted in the light of the lessons learned and interpretations already gathered under the GDPR. Furthermore, it will be analyzed how the main challenges for fundamental rights that are posed by such decision-making algorithms can be summarized under the principles of transparency, accountability and justice/equality. A multifaceted and multidisciplinary approach is proposed, to be applied at different moments in time, to ensure that such principles are incorporated during the development and use of machine learning decision-making algorithms. Finally, this thesis proposed that guaranteeing a right to explanation, which is currently allocated in a broader discussion involving accountability, must take into account a perspective of merit and procedure. The different types of content that have been mapped as likely to be required as an explanation are identified, as well as the values and rights that a right to explanation aims to protect, demonstrating, finally, the importance that such content be subject to public scrutiny.

The aim of this qualitative interview study is to describe and analyze social workers experiences of the measures to prevent client related violence in the Swedish child welfare with focus on protection of personal data as a specific preventive measure of client related violence. Data from six social workers was analyzed by means of a thematic analysis: The research show that the social workers experienced occurrence of client related violence in the Swedish child welfare is low due to defaulted definition, tolerance, and normalization of client related violence in the organizational culture. The measures to prevent client related violence is based on governmental routines and guidelines as well as the social workers' own knowledge about and ability to manage violence. An overall conclusion of the study is that social workers find their name as a central part of their profession. As a result of this, the study shows that the specific preventive measure protection of personal data is problematic partly due to the risk to impede the relationship with clients’ and partly due to the risk of disregarding clients’ right to legal certainty. Consequently, there is a need for other preventive measures.

Исследование возможности использования пользовательской информации с целью извлечения прибыли в связи с развитием технологий является актуальным для экономической сферы деятельности любой отрасли коммерческого предприятия. Первой задачей данной диссертационной работы является разработка и описание модели, предназначенной для эффективного ее использования с целью монетизации результатов анализа персональных данных, с учетом выявленных в процессе исследования недостатков существующих решений, применяемых на рынке. Второй задачей работы является разработка рабочей модели обработки данных, лежащей в основе предложенной модели монетизации с последующим проведением анализа точности ее работы с целью подтверждения жизнеспособности идеи разработки полноценной предложенной модели с целью решения поставленных перед ней задач. / Study the possibility of using user information for profit in connection with the development of technology is relevant to the economic activity of any branch of business. The aim of this work is the development and description of models, designed for efficient use with the aim of monetizing the results of the analysis of personal data, as well as develop a working model of processing underlying the proposed model of monetization. Here will be described the alternative methods and technologies that can use to build similar systems. Practical aspect in this regard is the lack of automated systems to determine the personality type of the person based on his activity on the Internet. Including descriptions of the use of such systems to resolve problems in different areas on the basis of an assessment of the accuracy of their work.

Тема магистерской диссертации: Применение искусственного интеллекта при обработке анкетных данных. Магистерская диссертация выполнена на 98 страницах, содержит 13 таблиц, 30 рисунков, 62 использованных источника. Актуальность темы обусловлена большими трудозатратами и нерелевантными результатами обработки анкетных данных. Целью работы является автоматизация процесса отбора анкетных данных в дистрибутиве Python Anaconda с использованием алгоритмов машинного обучения. Задачи работы:  изучить системы искусственного интеллекта;  рассмотреть программное обеспечение для систем искусственного интеллекта;  создать и обучить классификатор для сортировки анкетных данных;  оценить экономическую эффективность создания проекта. Объект исследования  система сбора и обработки анкетных данных отдела диспетчеризации ВШЭМ УрФУ. Предмет исследования  автоматизация процесса ранжирования анкетных данных по релевантности. В первой главе рассматривается обработка данных с использованием систем искусственного интеллекта. Вторая глава посвящена разработке методики использования систем искусственного интеллекта при обработке анкетных данных. В третьей главе представлены системы искусственного интеллекта при сборе и обработке анкетных данных Результаты работы: практическим результатом работы стал разработанный классификатор, который определяет для заполненной анкеты: будет ли она учтена для анализа эффективности учебного процесса. / Theme of the master's thesis: Application of artificial intelligence in the processing of personal data. The master's thesis is done on 98 pages, contains of 13 tables, 30 figures, 62 literature sources. The relevance of the topic is due to the high labor costs and irrelevant results of the personal data processing. The purpose of the work is to automate the process of selecting personal data in the Python Anaconda distribution using machine learning algorithms. Objectives of work:  to explore artificial intelligence systems;  to consider software for artificial intelligence systems;  to create and train a classifier for the personal data sorting;  to evaluate the economic effectiveness of the project. The object of the study is the system for personal data collecting and processing of the dispatch department of the Higher School of Economics of UrFU. The subject of the research is the automation of the process of ranking the questionnaire data by relevance. The first chapter deals with the processing of data using artificial intelligence systems. The second chapter is devoted to the development of methods for the use of artificial intelligence systems in the processing of personal data. The third chapter presents artificial intelligence systems for the collection and processing of personal data The results of the work: the practical result of the work was the developed classifier, which defines for the completed questionnaire: it would be taken into account for impact analysis of the educational process.

[pt] É possível falar em uma autogestão plena sobre os próprios dados na internet? Esta pergunta, norteadora da presente dissertação, orienta o propósito deste estudo, que tem por objetivo verificar se, em termos práticos, os titulares de dados detêm efetivo controle quanto ao uso de suas informações pessoais no ciberespaço. Sob a análise de diferentes arranjos entre relações on-line e dos respectivos meios de ingerência disponibilizados ao titular, busca-se traçar parâmetros condizentes ao autogoverno sobre os dados, em atenção aos pressupostos do direito à autodeterminação informativa, basilar à sistemática de proteção de dados pessoais. De modo a identificar essas circunstâncias fáticas, a pesquisa traz como enfoque a contextualização do consentimento, enquanto possível hipótese legal capaz de autorizar o tratamento de dados. Acredita-se que uma maior compreensão quanto a formas devidas de controle informacional exercido pelos titulares será determinante à construção de relações de tratamento mais íntegras e confiáveis na internet. / [en] Is it possible to say that there is a full self-management of one s own data on the internet? This question, which guides this present dissertation, guides the purpose of this study, which aims to verify whether, in practical terms, data subjects have effective control over the use of their personal information in cyberspace. Under the analysis of different arrangements between online relations and their respective means of interference made available to the data subject, an attempt is made to outline parameters consistent with self-government over data, considering the assumptions of the right to informational self-determination, basic to the personal data protection system. In order to identify these factual circumstances, the research focuses on the contextualization of consent, as a possible legal hypothesis capable of authorizing data processing. It is believed that a greater understanding of the proper forms of informational control exercised by data subjects will be crucial to the construction of a more complete and reliable processing relations on the internet.

[pt] O presente trabalho pretende analisar como a desinformação é monetizada no ambiente virtual e compreender aspectos regulatórios estruturais e concretos para reduzir os efeitos nocivos desta prática. Assim, será demostrado o conceito de desinformação e de publicidade personalizada, além da relação que possuem entre si. Em virtude dos diversos benefícios gerados por este tipo de publicidade e dos riscos para a liberdade de expressão que podem advir da regulação focada em conteúdo, são apresentadas soluções centradas na arquitetura da plataforma e na proteção de dados pessoais. Como proposta, se apresenta uma regulação multiparticipativa, com amplitude de instrumentos e com a mescla de conceitos mais precisos e mais vagos, buscando segurança jurídica sem descuidar da necessária elasticidade diante da dinâmica que envolve novas tecnologias. Por fim, são apresentadas medidas concretas voltadas para (i) transparência e empoderamento do usuário; (ii) transparência e controle para o anunciante; (iii) accountability e dados pessoais. / [en] The present work intends to analyze how disinformation is monetized on the virtual environment and comprehend the concrete and structural regulatory aspects to reduce the damaging effects of said practice. Thus, the concept of disinformation and programmatic advertising will be shown, as well as the link between them. By virtue of many benefits gerated by this kind of advertising and the risks to the freedom of speech that can come from regulation focussed on contente, solutions based on the the architecture of the plataform and personal data privice protection are presented. As a proposal, a multi-stakeholder regulation is presented with te amplitude of mechanisms and with the mix of the most accurate and vague concepts looking for legal security without neglecting the elasticity there is required in view of the dynamics that involves new thecnologies. Finally, concrete measures designed for (ii) transparency andu ser empowerment are presented; (ii) transparency and control for the advertiser; (iii) accountability and personal data.

In this thesis, a novel approach to classifying personal data is tested. Previous personal data classification models read the personal data before classifying it. However, this thesis instead investigates an approach to classify personal data by looking at contextual information frequently available in data sets. The thesis compares the well-researched word embedding methods Word2Vec, Global representations of Vectors (GloVe) and Bidirectional Encoder Representations from Transformers (BERT) used in conjunction with the different types of classification methods Bag Of Word representation (BOW), Convolutional Neural Networks (CNN), and Long Short-term Memory (LSTM) when solving a personal data classification task. The comparisons are made by extrinsically evaluating the different embeddings' and models' performance in a personal data classification task on a sizable collection of well-labeled datasets belonging to Spotify. The results suggest that the embedded representations of the contextual data capture enough information to be able to classify personal data both when classifying non-personal data against personal data, and also when classifying different types of personal data from each other. / I denna uppsats undersöks ett nytt tillvägagångssätt att klassificera personlig data. Tidigare dataklassificerings modeller läser data innan den klassificerar den. I denna uppsats undersöks istället ett tillvägagångssätt där den kontextuella informationen används. Uppsatsen jämför flera väletablerade metoder för 'word embedding' så som Word2Vec, Global representations of Vectors (GloVe) och Bidirectional Encoder Representations from Transformers (BERT) i kombination med klassificeringsmodellerna Bag Of Word representation (BOW), Convolutional Neural Networks (CNN) och Long Short-term Memory (LSTM). Modellerna jämförs genom att evaluera deras förmåga att klassificera olika typer av personlig data baserad på namngivning och beskrivning av dataset. Resultaten pekar på att representationerna samt modellerna fångar tillräckligt med information för att kunna klassificera personlig data baserat på den kontextuell information som gavs. Utöver detta antyder resultaten att modellerna även klarar av att urskilja olika typer av personlig data från varandra.

With the exponential proliferation of data, there has been a surge in data generation fromdiverse sources, including social media platforms, websites, mobile devices, and sensors.However, not all data is readily visible or accessible to the public, leading to the emergence ofthe concept known as "dark data." This type of data can exist in structured or unstructuredformats and can be stored in various repositories, such as databases, log files, and backups.The reasons behind data being classified as "dark" can vary, encompassing factors such as limited awareness, insufficient resources or tools for data analysis, or a perception ofirrelevance to current business operations. This research employs a qualitative research methodology incorporating audio/videorecordings and personal interviews to gather data, aiming to gain insights into individuals'understanding of the risks associated with dark data and their behaviors concerning thesharing of personal information online. Through the thematic analysis of the collected data,patterns and trends in individuals' risk perceptions regarding dark data become evident. The findings of this study illuminate the multiple dimensions of individuals' risk perceptions andt heir influence on attitudes towards sharing personal information in online contexts. Theseinsights provide valuable understanding of the factors that shape individuals' decisionsconcerning data privacy and security in the digital era. By contributing to the existing body ofknowledge, this research offers a deeper comprehension of the interplay between dark datarisks, individuals' perceptions, and their behaviors pertaining to online information sharing.The implications of this study can inform the development of strategies and interventionsaimed at fostering informed decision-making and ensuring personal safety in an increasinglydata-centric world