Taffarel Brant-Ribeiro

An Automatic and Dynamic Student Modeling Approach for Adaptive and Intelligent Educational Systems Using Ontologies and Bayesian Networks

Dynamic adaptation of educational content has been an important research topic. Therefore, in order for it to run effectively, student models that properly describe and monitor the cognitive state of students are needed. In this sense, this paper presents a hybrid student model approach that combines ontologies and Bayesian Networks to identify the knowledge of each student based on their characteristics and behavior while using an Adaptive Educational System. Experiments were performed with real student participants in a higher education course using an experimental prototype developed to verify the viability of the approach, which showed satisfactory results.

Interactive web interfaces modeling, simulation and analysis using Colored Petri Nets

Interaction modeling is a relevant activity during software development processes. Created relying on Petri Nets theory and aiming to represent discrete time events, Colored Petri Nets (CPNs) are a graphical formal language developed and widely employed for system modeling. While traditional CPNs only have elements with ordinary stylization and behaviors, in this article we explore the key ideas behind Web Interaction Modeling Using Colored Petri Nets (wiCPN), a modeling style developed with focus on representing Web interactions as an incremental improvement of CPNs. We review wiCPN’s refinements over CPNs and the modeling of the Web interface of Classroom eXperience (CX), a ubiquitous educational platform, thus verifying the model’s properties to ensure it was able to represent the different access levels among its users and how wiCPN displayed suitability to comprehend this requirement on the generated model. We have also improved the originally developed model with the modification of elements to make it finite and fully analyzable. Also, we added temporization capabilities to the model and ran corresponding user simulation to observe the average time that users with different roles tend to spend during interactions. We compared wiCPN results with Unified Modeling Language (UML) Activity and Use Case diagrams, observing, as outcomes, that the generated model represented CX’s interactive flow correctly and maintained a concise notation—a single wiCPN diagram was sufficient to depict the same interactive flow that, in UML, would require several diagrams, something that could overload the design team in actual software development scenarios. We also included new user experiments comprising qualitative results from experts. Finally, we created a reachability graph for the new model and generated a full state space report, analyzing Petri Nets properties such as boundedness, liveness and home marking.

An Automatic and Dynamic Knowledge Assessment Module for Adaptive Educational Systems

Adaptive Educational Systems (AES) make use of Artificial Intelligence techniques aiming at adapting themselves to the real needs of the student, and through such provide a personalized and individualized teaching. In order for this adaptation to be successful, it is important that the system knows the level of knowledge concerning the real cognitive state of the students. In this manner, this article presents an approach for predicting academic performance based on ontologies and Bayesian networks. A knowledge assessment module was developed and integrated to a Ubiquitous Learning Environment (ULE) in real and actual use. During three semesters, five graduate classes were analyzed with the intent of verifying the correlation between the grades generated by the module and the actual grades obtained by the students. It was noted that 80% of the analyzed samples obtained very high or substantial correlations. This indicates that the proposed module is adequate for identifying and predicting academic performance.