Research shows that expert-crafted worked examples can have a positive effect on student performance. To investigate the potential for data-driven worked examples to achieve similar results, we generated worked examples for the Deep Thought logic tutor, and conducted an experiment to assess their impact on performance. Students who received data-driven worked examples were much more likely to c...

We have augmented the Deep Thought logic tutor with a Hint Factory that generates data-driven, context-specific hints for an existing computer aided instructional tool. We investigate the impact of the Hint Factory’s automatically generated hints on educational outcomes in a switching replications experiment that shows that hints help students persist in a deductive logic proofs tutor. Three in...

Many tutors offer students reference material or tips that they can access as needed. We have logged data about student use of references with Deep Thought logic tutor which to understand why and how references are used. We find evidence that students use these references in systematic ways that change over the course of the tutor, and can be predictive of rule application errors. We can use th...

We propose a novel extension of the encoder-decoder framework, called a review network. The review network is generic and can enhance any existing encoderdecoder model: in this paper, we consider RNN decoders with both CNN and RNN encoders. The review network performs a number of review steps with attention mechanism on the encoder hidden states, and outputs a thought vector after each review s...

structures

We have proposed a novel application of Markov decision processes (MDPs), a reinforcement learning technique, to automatically generate hints using historical student data. Using this technique, we have modified a an existing, non-adaptive logic proof tutor called Deep Thought with a Hint Factory that provides hints on the next step a student might take. This paper presents the results of our p...

This work explores the effects of using automatically generated hints in Deep Thought, a propositional logic tutor. Generating hints automatically removes a large amount of development time for new tutors, and it also useful for already existing computer-aided instruction systems that lack intelligent feedback. We focus on a series of problems, after which, the control group is known to be 3.5 ...

The interactions of concepts and problem-solving techniques needed to solve open-ended proof problems are varied, making it difficult to select problems that improve individual student performance. We have developed a system of datadriven ordered problem selection for Deep Thought, a logic proof tutor. The problem selection system presents problem sets of expert-determined higher or lower diffi...

Editor’s note: This editorial is part of a series written by editors and co-authored with a senior executive, thought leader, or scholar from a different field to explore new content areas and grand challenges with the goal of expanding the scope, interestingness, and relevance of the work presented in the Academy of Management Journal. The principle is to use the editorial notes as “stage sett...

Automatic problem generation for learning tools can provide the required quantity and variation of problems necessary for an intelligent tutoring system. However, this requires an understanding of problem difficulty and corresponding features of student performance. Our goal is to automatically generate new proof problems in Deep Thought – an online propositional logic learning tool – for indiv...