Analysing Problem Sequencing Strategies Based on Revised Bloom's Taxonomy Using Deep Knowledge Tracing

摘要

Revised Bloom's Taxonomy (RBT) is hierarchical in nature and it serves as a common vocabulary for the teachers to classify learning objectives of a curriculum. In this work, we study the effects of using RBT as a problem sequencing strategy on students' learning. We compare a blocking strategy based on RBT against the random strategy. We also implement the reversed hierarchical order of this taxonomy as a strategy to understand the effect of a contrast behaviour, if any. We also examine both forward and reverse hierarchical orders by enhancing them with interleaving behaviour. We use deep learning based knowledge tracing model, Deep Knowledge Tracing to simulate the students' behaviour. We observe that forward hierarchical order yields a significant gain over reverse hierarchical order. Interestingly, interleaving on RBT did not outperform blocking strategy as expected [6].