PPGI - Dissertações de mestrado

URI Permanente para esta coleção


Submissões Recentes

Agora exibindo 1 - 5 de 215
  • Item
    Ontological foundations for conceptual modeling datatypes
    (Universidade Federal do Espírito Santo, 2013-08-28) Albuquerque, Antognoni Fundão de; Guizzardi, Giancarlo; Barcellos, Monalessa Perini; Cruz, Fernando
  • Item
    TOM: tutor inteligente orientado a construção de mapas conceituais
    (Universidade Federal do Espírito Santo, 2019-02-07) Boguski, Rodrigo Ruy; Cury, Davidson; Pinto, Sérgio Crespo Coelho da Silva; Menezes, Crediné Silva de
    Tutoring is a method for pedagogical interaction used for thousands of years. In this interaction, the tutor evaluates the student's learning during the course of the learning experiences, heals his doubts, follows his frequency, assists his motivation and provides support, usually in a timely manner, so that the cognitive overload resulting from the learning process itself is minimized, order to make that moment pleasant and satisfying. Having the knowledge of the needs of the students is still a great challenge for the teachers, because this is the first step to providing relevant help in order to contribute to their learning. From this perspective we develop a technique based on rules of association of data mining in order to identify the conceptual flaws in students so that, once this knowledge gap is known, it is filled in order to potentialize the cognitive process. In order to test these rules, we constructed an intelligent tutor system capable of using them and thus guide the student in the construction of conceptual maps from a proposed theme, having as main guideline the teacher's reference map and the respective rules of association that each concept of this map has. Thus, the missing information to the student is not presented in a destructured way, but with different levels of granularity, similarities and in the most appropriate sequence to catalyze learning. This method has as its underlying question a pedagogical approach that uses aspects of the theory of meaningful learning defended by Ausubel and complemented by Novak.
  • Item
    O2CMF: um framework para experimentação federada em NFV
    (Universidade Federal do Espírito Santo, 2019-02-14) Ceravolo, Isabella de Albuquerque; Martinello, Magnos; Rezende, José Ferreira de; Mota, Vinicius Fernandes Soares
    Federated testbeds support network research by providing a distributed lab. This is possible through frameworks that transform physical resources in an experimentation service. Such service needs to continuously evolve including emerging technologies. Network Function Virtualization (NFV) is an emerging technology which enables to decouple network intelligence from physical hardware. Although frameworks (such as GCF, OCF, and OMF) have strongly contributed to the establishment of network testbeds’ federations, they lack features required to support NFV. This is due to the type of virtualization, monitoring, and resources which they offer. Besides that, they lack NFV orchestration functionalities. This work proposes a new framework to enable NFV experimentation. The result was O2CMF, a framework based on the OpenStack cloud computing platform and interoperable with the Fed4FIRE infrastructure. To validate O2CMF, we developed demonstrations showing testbed management, Fed4FIRE compatibility, traffic isolation, NFV orchestration and integration with other domains (robotics, wireless networks, and OpenFlow). These proofs of concept, we demonstrated that O2CMF successfully enabled federated experimentation in NFV, combining the interoperability provided by SFA with the flexibility and robustness of the cloud, and orchestration features. O2CMF have been used to manage a testbed at UFES, supporting research and education activities. In addition, its documentation, covering operation and use, motivated its adoption by the University of Bristol.
  • Item
    ProScene: uma plataforma para simulação de situações
    (Universidade Federal do Espírito Santo, 2018-11-20) Baldi, Alessandro Murta; Costa, Patrícia Dockhorn; Almeida, João Paulo Andrade
    Systems simulation permits that processes in real life to be developed in a controlled environment allowing experimentation over a wide range of conditions. These experiments promote (i) the identification of possible problems that occur over time in a given context, (ii) the training of simulated systems and organisms and (iii) analyzes of possible states that systems can reach. This dissertation explores an aspect not yet found in the literature: the use of simulations that explicitly consider the concept of Situation, named in this work as Situation-Aware Simulations (SiSA). Therefore, SiSAs are designed to simulate the execution of situationbased systems, capable of adapting autonomously to the situation of its users, promoting an innovative form of feedback, close to what happens in reality. The purpose of this work is to facilitate the development of SiSAs and, in this sense, it provides two important contributions: (i) an exploratory research in several simulation tools comparing programming paradigms, performance and characteristics of tools for the development of a SISA, and (ii) new simulations and situations platform called ProScene. ProScene is a hybrid platform, with the features of an agent-oriented simulation tool and a situation management platform. In this way, ProScene has a unique characteristic: it enables the developer to implement situations as agents of the simulation, allowing the visual monitoring of activation and deactivation of situations in their locations.
  • Item
    Proposta de um sistema automático de avaliação de redações do ENEM, foco na competência 1: demonstrar domínio da modalidade escrita formal da língua portuguesa
    (Universidade Federal do Espírito Santo, 2017-08-03) Almeida Júnior, Celso Romão Cardoso de; Oliveira, Elias Silva de; Lima, Priscila Machado Vieira; Cury, Davidson
    Automatic essay assessments are widely practiced in English, but in Portuguese, we cannot say the same. Writing is one of the competencies required by the National High School Examination (ENEM), gateway to most universities in Brazil. The high cost and the large number of professionals working in the correction process of the ENEM essays are some of the factors that motivate research in the area of automatic essay evaluation. This work presents a strategy to improve the evaluator’s productivity, reducing the effort in 20% of the time, evaluating Competence 1, one of the five competences evaluated in the essays of ENEM. For this, we propose the construction of a system of automatic evaluation of essays, in Competence 1 of the ENEM; demonstrate mastery of the formal written form of the Portuguese language. In the construction of the system, we use Natural Language Processing techniques and tools, in the preprocessing stage of the essays as well as Machine Learning techniques in the stages of selection of characteristics and prediction of the grades. The results of the experiments carried out with the UOL website show that the system is able to support the ENEM essay evaluator with an absolute mean error of 0.2354 in 2.0 compared to the scores attributed by the site experts.