The acquisition of German declension in additive and concept-based approaches to instruction via computer-based cognitive tutors

This study examines the different learning outcomes of two computer-based cognitive tutors using two approaches to instructing German declension: an additive, bottom-up approach, which focuses on a stepwise introduction of each case, and a concept-based, top-down approach, which focuses on developing students’ conceptual understanding of the functions related to each case form and the case marking system as a whole. The results indicate that both groups learned, but what and how they learned differed depending on the method of instruction. The additive group showed general gains in production and a slight increase in their ability to correctly interpret object-first sentences. The concept-based group showed larger gains, but in fewer areas. Specifically, the production of adjective endings increased, although there were no differences in determiner production or accuracy. The concept-based group also had a larger gain in their ability to interpret non-SVO word order sentences. This study shows how concept-based approaches to grammar can outperform additive ones, and that the development of these concepts can prepare students for future learning.

The 2010 KDD Cup Competition Dataset: Engaging the machine learning community in predictive learning analytics

In the spring of 2010, the Association for Computing Machinery (ACM) Special Interest Group on Knowledge Discovery and Data-mining (KDD) selected a dataset from an educational technology for its annual competition. The competition, titled “Educational Data Mining Challenge”, tasked participants with predicting the correctness of student answers to questions within an Intelligent Tutoring System (ITS) from The Cognitive Tutors suite of tutors. This challenge was hosted by the PSLC DataShop, and included data provided by the Carnegie Learning Inc., producers of The Cognitive Tutors. Consisting of over 9GB of student data this was the largest KDD Cup dataset up to that point in time. The competition brought in 655 competitors submitting 3,400 solutions. Five years later, we believe the competition dataset has been the most often cited from an educational technology platform.