Intelligent tutors are becoming more popular with the increased use of computersand hand held devices in the education sphere. An area of research isinvestigating how machine learning can be used to improve the precision andfeedback of the tutor. This thesis compares machine learning clustering algorithmswith various distance functions in an attempt to cluster together codesnapshots of students solving a programming task. It investigates whethera general non-problem specific implementation of a distance function canbe used to identify when a student is stuck solving an assignment. Themachine learning algorithms compared are k-medoids, the randomly initializedalgorithm that produces a pre-defined number of clusters and affinitypropagation, a two phase algorithm with dynamic cluster sizes. Distancefunctions tried are based on the Bag of Words approach, lower level APIcalls and a problem specific distance function. This thesis could not find agood algorithm to achieve the sought goal, and lists a number of possibleerror sources linked to the data, preprocessing and algorithm. The methodologyis promising but requires a controlled environment at every level toassure data quality does not detract from the analysis in later stages.

This study investigated the various methods involved in creating an intelligent tutor for the University of Central Florida Infinity Web Applets (UCF Infinity Web Applets). After conducting research into various methods, two major methods emerged and they are: solving the problem for the student and helping the student when they become stymied and unable to solve the problem. A storyboard was created to show the interactions of the student and system along with a list of features that were desired to be included in the tutoring system. From the storyboard and list of features, an architecture was created to handle all of the interactions and features. After the initial architecture was designed, the development of the actual system was started. The architecture underwent a multitude of changes to conclude with a working system, EINO. The final architecture of EINO incorporated a case based reasoning system to perform pattern recognition on the student's input into the UCF Infinity Web Applets. The interface that the student interacts with was created using flash. EINO was implemented in three of the labs from the UCF Infinity Web Applets. A series of tests were performed on the EINO tutoring system to prove that the system could actually perform each and every one of the features listed initially. The final test was a simulation of how the EINO would perform under a set of given cases. Test subjects with the same educational level as the target group were chosen to spend an unlimited time using each of the three labs. Each of the test subjects filled out a survey on every lab to determine if the EINO system produced a helpful output.

abstract: Research has shown that the learning processes can be enriched and enhanced with the presence of affective interventions. The goal of this dissertation was to design, implement, and evaluate an affective agent that provides affective support in real-time in order to enrich the student’s learning experience and performance by inducing and/or maintaining a productive learning path. This work combined research and best practices from affective computing, intelligent tutoring systems, and educational technology to address the design and implementation of an affective agent and corresponding pedagogical interventions. It included the incorporation of the affective agent into an Exploratory Learning Environment (ELE) adapted for this research. A gendered, three-dimensional, animated, human-like character accompanied by text- and speech-based dialogue visually represented the proposed affective agent. The agent’s pedagogical interventions considered inputs from the ELE (interface, model building, and performance events) and from the user (emotional and cognitive events). The user’s emotional events captured by biometric sensors and processed by a decision-level fusion algorithm for a multimodal system in combination with the events from the ELE informed the production-rule-based behavior engine to define and trigger pedagogical interventions. The pedagogical interventions were focused on affective dimensions and occurred in the form of affective dialogue prompts and animations. An experiment was conducted to assess the impact of the affective agent, Hope, on the student’s learning experience and performance. In terms of the student’s learning experience, the effect of the agent was analyzed in four components: perception of the instructional material, perception of the usefulness of the agent, ELE usability, and the affective responses from the agent triggered by the student’s affective states. Additionally, in terms of the student’s performance, the effect of the agent was analyzed in five components: tasks completed, time spent solving a task, planning time while solving a task, usage of the provided help, and attempts to successfully complete a task. The findings from the experiment did not provide the anticipated results related to the effect of the agent; however, the results provided insights to improve diverse components in the design of affective agents as well as for the design of the behavior engines and algorithms to detect, represent, and handle affective information. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2018

Dans cette thèse, nous présentons MEMORIA, un modèle de représentation de la mémoire de l’apprenant pour les systèmes tutoriels intelligents et adaptatifs. La contribution principale de ce modèle est une formalisation et une implémentation du modèle de l’apprenant sous forme de mémoires qui stockent les informations perçues par l’apprenant dans un environnement virtuel et les instructions émises par le tuteur. La conception de notre modèle est basée sur les quatre composantes classiques d’un système tutoriel intelligent. Le modèle du domaine est représenté par les connaissances métiers formalisées à l'aide de MASCARET. Afin de rendre naturelles les interactions entre le tuteur et l'apprenant, nous représentons le modèle de l'interface par l'intermédiaire d'un agent conversationnel animé à l'aide de la plate-forme Greta. Le modèle de l'apprenant est constitué de l'ensemble des connaissances acquises par l'apprenant en cours de simulation. Ces connaissances sont organisées dans trois mémoires : la mémoire sensorielle, la mémoire de travail et la mémoire à long terme. Notre enjeu majeur porte sur la formalisation de l'encodage des informations dans ces mémoires, ainsi que le flux de données entre celles-ci. Cette formalisation est basée sur la théorie de la mémoire humaine proposée par Atkinson et Shiffrin et inspirée de l'architecture cognitive ACT-R. Le modèle de tuteur que nous proposons est centré sur la réalisation d'un comportement qui adapte l'exécution du scénario en fonction des connaissances de l'apprenant et de ses interactions avec le tuteur. Une étude expérimentale a été menée pour valider notre modèle. Nous avons comparé deux groupes de participants. Dans le premier groupe, nous avons intégré un tuteur adaptatif utilisant notre modèle, qui adapte l'exécution du scénario pédagogique et dans le second groupe, un tuteur non adaptatif qui réalise un scénario pédagogique figé. Les résultats de cette étude permettent de conclure quant à l'efficacité de notre modèle pour un apprentissage de procédure. / In this thesis, we present MEMORIA, a model of the learner’s memory representation for adaptive and intelligent tutoring systems. The main contribution of this model is a formalization and an implementation of the learner's model using memories that store the information perceived by the learner in a virtual environment and the instructions given by the tutor. The design of our model is based on the four classic components of an intelligent tutorial system.The domain model is represented by the domain knowledge that is formalized using MASCARET. In order to make the interactions between the tutor and the learner natural, we represent the interface model through an embodied conversational agent using GRETA. The learner's model is made of all the knowledge acquired by the learner during the simulation. This knowledge is organized into three memories: sensory memory, working memory, and long-term memory. Our major challenge is to formalize the encoding of information in these memories, as well as the data flow between them.This formalization is based on the theory of human memory proposed by Atkinson and Shiffrin and inspired by the cognitive architecture ACT-R. Our proposed tutor model focuses on the realization of a behavior that adapts the execution of the pedagogical scenario according to the learner's knowledge and the interactions with the tutor. An experimental study was conducted to validate our model. We compared two groups of participants. In the first group, we integrated an adaptive tutor using our model which adapts the execution of the pedagogical scenario and in the second group, a non-adaptive tutor who applied a fixed pedagogical scenario. The results of this study allow us to conclude on the effectiveness of our model for procedural learning.

This work presents an interactive environment for learning about fractions, with mechanisms to support cooperative and adaptive interactions offered by tutors agents to human learners, focusing mainly on activities to solve problems. For this purpose, an architecture based on software agents and semantic Web services was proposed, therefore, we verify the functional viability of the proposal and, posteriorly, to present a revision of that architecture to suply some requirements not previously covered, beyond models that support to those interactions. With respect to interactions, the learner will receive support from both a pedagogical agent tutor, as some of their peers who are part of the environment. Particularly, a tutor agent has an open learner model, from which it obtains information to guide their actions. The idea of this model be opened is to allow the learner seeing the evaluation that the system has about him, and also the opportunity to disagree with this assessment, and thus contribute to the refinement of the content of such a model / Fundação de Amparo a Pesquisa do Estado de Alagoas / Este trabalho apresenta um ambiente interativo de aprendizagem sobre Frações, dotado de mecanismos de suporte a interações cooperativas e adaptativas oferecidas por seus agentes tutores aos aprendizes humanos, focando principalmente em atividades de resolução de problemas. Para isso, propõe-se uma arquitetura baseada em agentes de software e serviços Web semânticos, daí, pôde-se verificar a viabilidade funcional da proposta e, posteriormente, apresentar uma revisão de tal arquitetura para suprir alguns requisitos anteriormente não visados, além de modelos que dão suporte às referidas interações. No que diz respeito às interações, o aprendiz receberá suporte pedagógico tanto de um agente tutor, quanto de algum de seus pares que fazem parte do ambiente. Particularmente, um agente tutor conta com um modelo aberto do aprendiz, a partir do qual passa a dispor de informações úteis para orientar suas ações. A idéia deste modelo ser aberto é a de permitir que o aprendiz possa ver qual a avaliação que o sistema tem a seu respeito, tendo ainda a oportunidade de discordar de tal avaliação, e assim contribuir para o refinamento do conteúdo de tal modelo. Palavras-chave: Modelagem Aberta do Aprendiz; Sistemas Tutores Inteligentes; Sistemas Multi-agentes

This study is a qualitative research project classified as practical and participatory action research designs, which has as its aim the creation of an instructional design, used as a learning metric for Domus Software - Procel Edification, which integrates concept maps and the revised Bloom taxonomy into an intelligent web assessment system. This metric aligns curriculum concept maps along with conceptual and procedural knowledge of the Domus software - Procel edifies the retaining cognitive processes, provides understanding and application, through offering a model of instructional design, which assigns weights to those cognitive processes attained by students and identifies principles to be used in its applicability for the evaluation of distance learning. The authors also present the results for alignment, inferring weights as well as an outline of the logical sequence along with steps for the implementation of the intelligent system through the association of some exemplification slides. / O presente trabalho é uma pesquisa qualitativa e classificada como practical and participatory action research designs (desenho de pesquisa de prática e ação participativa), que tem como intuito criar um desenho instrucional para uma métrica da aprendizagem do Software Domus Procel Edifica, integrando mapas conceituais à taxonomia revisada de Bloom em um sistema inteligente de avaliação na Web. Essa métrica alinha mapas conceituais curriculares, conhecimentos procedimentais e conceituais do software Domus − Procel Edifica aos processos cognitivos de retenção, entendimento e aplicação, oferecendo um modelo de desenho instrucional, que atribui pesos aos processos cognitivos alcançados pelos estudantes e identifica alguns princípios para sua aplicabilidade na avaliação da aprendizagem a distância. Apresenta-se ainda, resultados de alinhamento, inferência de pesos e um esboço da sequência lógica e etapas de execução do sistema inteligente, associando-se algumas telas de exemplificação. / Mestre em Educação