No abstract

Technology-based, open-ended learning environments (OELEs) can capture detailed information of students' interactions as they work through a task or solve a problem embedded in the environment. This information, in the form of log data, has the potential to provide important insights about the practices adopted by students for scientific inquiry and problem solving. How to parse and analyse the log data to reveal evidence of multifaceted constructs like inquiry and problem solving holds the key to making interactive learning environments useful for assessing students' higher-order competencies. In this paper, we present a systematic review of studies that used log data generated in OELEs to describe, model and assess scientific inquiry and problem solving. We identify and analyse 70 conference proceedings and journal papers published between 2012 and 2021. Our results reveal large variations in OELE and task characteristics, approaches used to extract features from log data and interpretation models used to link features to target constructs. While the educational data mining and learning analytics communities have made progress in leveraging log data to model inquiry and problem solving, multiple barriers still exist to hamper the production of representative, reproducible

Interactive simulations allow students to discover the underlying principles of a scientific phenomenon through their own exploration. Unfortunately, students often struggle to learn effectively in these environments. Classifying students' interaction data in the simulations based on their expected performance has the potential to enable adaptive guidance and consequently improve students' learning. Previous research in this field has mainly focused on a-posteriori analyses or investigations limited to one specific predictive model and simulation. In this paper, we investigate the quality and generalisability of models for an early prediction of conceptual understanding based on clickstream data of students across interactive simulations. We first measure the students' conceptual understanding through their in-task performance. Then, we suggest a novel type of features that, starting from clickstream data, encodes both the state of the simulation and the action performed by the student. We finally propose to feed these features into GRU-based models, with and without attention, for prediction. Experiments on two different simulations and with two different populations show that our proposed models outperform shallow learning baselines and better generalise to different learning environments and populations. The inclusion of attention into the model increases interpretability in terms of effective inquiry. The source code is available on Github 1 .

No abstract