http://hdl.handle.net/11422/88 Tue, 07 Apr 2026 21:54:21 GMT 2026-04-07T21:54:21Z http://hdl.handle.net/11422/28740 Title: Lógica Fuzzy aplicada à análise comportamental e conhecimento da guerra cognitiva em redes sociais: um modelo de extração e mineração de dados Author(s)/Inventor(s): Ferreira, Vinícius Marques da Silva Advisor: Cosenza, Carlos Alberto Nunes Abstract: The research aims to understand how behavior, emotion and subjectivity participate in the construction of a conflict or polarization of ideas as an instrument of Cognitive Warfare in social networks, between actors or groups of actors, for example on the Twitter and Facebook platform, through the application of computational methods or models that work on subjectivity between them, using algorithms based on machine learning. The sentiment of the actors' dialogues on the subject addressed in terms of the polarity of the content, classifying opinions, for example whether they are positive or neutral or negative, is also analyzed. We will use text mining and sentiment analysis techniques. The aim is to develop and present a text processing and analysis system that optimizes the identification of polarizing ideas and improves multi-criteria decision making within diverse contexts on social networks, taking into account feelings and emotions. The approach consists of identifying and collecting the opinions of Twitter users on certain subjects, in the effervescence of events surrounding the issues listed. Text mining and sentiment analysis techniques are applied to extract qualifiers from the data analyzed. In addition, we developed a data visualization system that relies on the parameterization of value ranges for colors. Our data suggests what is already known in the light of psycholinguistics: when humans express themselves, they can reveal a variety of emotions that they experience internally and that inspire their subjective experience, their actions and their communication. Publisher: Universidade Federal do Rio de Janeiro Type: Tese Sat, 01 Jun 2024 00:00:00 GMT http://hdl.handle.net/11422/28740 2024-06-01T00:00:00Z http://hdl.handle.net/11422/28739 Title: Falhas, causas e efeitos na comunicação de massa radiofônica, um estudo de caso de Engenharia de Produção na Super Rádio Tupi do Rio de Janeiro: uma aplicação Fuzzy Author(s)/Inventor(s): Barbosa, Sílvio Advisor: Cosenza, Carlos Alberto Nunes Abstract: The mass communication industry is constantly transforming, directly affecting radio stations and requiring effective management to ensure their survival and competitiveness. This study, focused on Super Rádio Tupi FM in Rio de Janeiro, one of the most traditional and influential in the country, identified and prioritized 38 failure factors, considering digital media such as WhatsApp, Facebook, X (formerly Twitter), and YouTube. Based on the 8M tool, known as the Cause and Effect Diagram, and using a fuzzy model developed and applied with the participation of 19 experts with extensive experience in broadcasting, activities from pre-production to post-production were analyzed. This study not only expands knowledge on radio activities but also offers significant practical and academic contributions by introducing a fuzzy modeling approach to broadcasting services, particularly in terms of identifying and mitigating failures in this productive environment. Publisher: Universidade Federal do Rio de Janeiro Type: Tese Mon, 01 Apr 2024 00:00:00 GMT http://hdl.handle.net/11422/28739 2024-04-01T00:00:00Z http://hdl.handle.net/11422/28489 Title: Um novo algoritmo agregação temporal para resolver processos de decisão markovianos com custo médio Author(s)/Inventor(s): Alexandre, Rodrigo e Alvim Advisor: Ferreira Filho, Virgílio José Martins Abstract: This work introduces and proves the convergence of a new class of algorithms based on time aggregation to solve Markov decision problems with average cost. The new algorithms use a new partitioning scheme that divides the state space into multiple subsets comprised of frontier states and interior states. The frontier states give rise to a new subset that comprises the state space of the embedded Markov chain. In the first step, the algorithm sets an ad-hoc policy for interior states and iterates only on frontier states. In contrast, in the second step, the algorithm iterates only on the interior states. One of the great advantages of the proposed approach is that the new partitioning scheme generates subsets with specific communication properties that allow a distributed policy improvement in the interior states, thus limiting the computational effort. Different policy improvement strategies were implemented and evaluated using the proposed approach to solve two problems: one is a production and inventory management problem and the other is a queue management problem. The results are very encouraging, as the proposed algorithms converged up to 60(20) times faster than the policy(value) iteration in the experiments. In addition to the new algorithms, this work presents a set of conceptual maps that characterise the state-of-the-art in Markov decision processes, Publisher: Universidade Federal do Rio de Janeiro Type: Tese Mon, 01 Apr 2024 00:00:00 GMT http://hdl.handle.net/11422/28489 2024-04-01T00:00:00Z http://hdl.handle.net/11422/28479 Title: Análise multicritério para seleção de pessoal no Programa Fragatas Classe Tamandaré Author(s)/Inventor(s): Almeida, Marcos Vinicius Passarini de Advisor: Marujo, Lino Guimarães Abstract: The Brazilian Navy Strategic Planning (PEM 2040), aligned with the National Defense Strategy, foresees the modernization of its resources through several initiatives, one of the most important being what is conventionally called the Tamandaré Class Frigate Program. The program envisages the construction of four modern, highly technologically complex ships. Within this context, the Brazilian Navy must decide, among its workforce, which people will receive the training and obtain the necessary knowledge for the maintenance of the new Frigates, so that in the future they can multiply this knowledge to others involved in their maintenance. The personnel selection to be trained is a complex process, which involves several alternatives and criteria and, therefore, lacks tools that are capable of reducing their subjectivity, increasing their efficiency and transparency, thus helping decision-making regarding the appropriate choice of personnel to be trained. This work aims to apply a personnel selection tool based on Multicriteria Decision Support methods, through research with a qualitativequantitative approach, through a case study. The result obtained demonstrated that the application of the tool was able to provide an ordered list of candidates for training, ranging from the candidate most compliant to the least compliant with the established criteria. The study is relevant due to its unprecedented nature in the Military Organization under study and the possibility that it can be replicated in other future personnel selection processes in the Brazilian Navy and other public companies. Publisher: Universidade Federal do Rio de Janeiro Type: Dissertação Mon, 01 Apr 2024 00:00:00 GMT http://hdl.handle.net/11422/28479 2024-04-01T00:00:00Z