AlgoSolve: supporting subgoal learning through learnersourced microtasks

Abstract

Algorithmic problem-solving has enabled scalable learning opportunities for learning programming. However, novices often struggle to solve problems successfully, where one of the main reasons arises from their inability to develop solution plans before solving the problem. Subgoal learning has been shown effective in promoting learners' ability to develop more complete solution plans. However, subgoal learning is most effective when expert-crafted guidance is provided, which is not widely available in self-learning environments. To overcome this problem, we developed a learnersourcing workflow composed of two microtasks that can collect high-quality subgoal labels and provide guidance in subgoal learning using these as high-quality examples. We implemented the workflow into AlgoSolve, a prototypical interface that supports subgoal learning in algorithmic problem-solving. Results from a between-subjects study with 63 novices demonstrate that AlgoSolve successfully guides learners to create high-quality subgoal labels and develop more complete solution plans.