Design of Adaptive Systems for Computer-based Learning Ruth González Novillo, Pedro J. Muñoz-Merino, Carlos Delgado Kloos UNIVERSIDAD CARLOS III DE MADRID
User Modelling Skills Cognitive states Emotions Motivation Engagement Gaming the system
Example: Skill modelling Inference of the skills of a student: Knowledge Spaces Item Response Theory Bayesian networks Semantic based
KNOWLEDGE SPACES What does a student know? What is a student ready to know? Structured knowledge A limited number of knowledge states Different learning paths Los ks definen un árbol que muestra las relaciones entre ejercicios q muestra el orden en q deben realizarse
KNOWLEDGE SPACES Different learning paths Personalization
ITEM RESPONSE THEORY Only one ability One, two or three parameters models: Difficulty Slope Need of calibration Guess ITEM CHARACTERISTIC CURVE (ICC) Pregunta posible -> limites por qué Em¡n la literatura son valores típicos.
ITEM RESPONSE THEORY ITEM RESPONSE FUNCTION (IRF) Probability of answering correctly an item depending on the ability Local independence of items. Estimation of the student ability. Pregunta posible -> limites por qué Em¡n la literatura son valores típicos.
BAYESIAN NETWORKS Nodes are usually exercises or skills Each node has a Conditional Probability Table(CPT) that denotes the probability of doing correctly an exercise depending on the results of the parent nodes Bayes Theorem for making the inference Conditional independence
Bayesian Networks An event updates the network It is possible to calculate the different probabilities
Semantic based solutions More rich semantic relationships among contents, e.g. using ontologies
COMPARISON ITEM RESPONSE THEORY BAYESIAN NETWORKS KNOWLEDGE SPACES +There is no semantic information +There is no need to build a network +One single skill +A priori calibration of parameters BAYESIAN NETWORKS KNOWLEDGE SPACES + More semantic information + Several skills + More complexity +Need of making tags, making the structure and probability assignment +No need of calibration +Predefined knowledge structure +No hidden node
Model for Adaptation Rules Adaptive rules that are atomic, parametric, reusable, interchangable 1) P.J. Muñoz-Merino, C. Delgado Kloos, M. Muñoz-Organero, & A. Pardo, “A software engineering model for the development of adaptation rules and its application in a hinting adaptive e-learning system,” Computer Science and Information Systems, vol. 12, no. 1, pp. 203-231, 2015
ISCARE: Adaptive Competition system P.J. Muñoz-Merino, M. Fernández Molina, M. Muñoz-Organero, & C. Delgado Kloos (2012), “An adaptive and innovative question-driven competition-based intelligent tutoring system for learning,” Expert Systems with Applications, vol. 39, no. 8, pp. 6932-6948.