Continuous Hint Factory (CHF)¶

Element · All elements
Evidence · no claims cited

Description¶

The CHF is a hint-generation method extending the Hint Factory to "vast and sparsely populated state spaces". It works in three steps: embed past student data in the edit distance space; infer a hint policy there via Gaussian process prediction for structured data, producing a weighted average of reference states chosen in a probabilistically optimal sense; and transform this mixture back into a human-readable edit via pre-image identification. By averaging multiple reference solutions it avoids emulating individual stylistic particularities without relying on frequency information.

Design Implications¶

Context¶

Requirements¶

• An edit distance and some, possibly few, data samples of past successful students

Constraints¶

• Requires trace data of successful students; the article recommends restricting data to successful students and goal-directed intermediate solutions

Target Learners¶

• students in intelligent tutoring systems for multi-step tasks such as programming

Target Learning Goals¶

• personalized next-step hints that reflect generic solution steps

Affordances¶

• Edit Distance Space Embedding Theorem

Claims¶

Related Elements¶

Examples¶

• Select hint-training data from successful students using a goal-directedness criterion

Key Sources¶

• Benjamin Paaßen, Barbara Hammer, Thomas W. Price, Tiffany Barnes, Sebastian Gross, Niels Pinkwart. (2018). The Continuous Hint Factory - Providing Hints in Vast and Sparsely Populated Edit Distance Spaces. Journal of Educational Data Mining, Volume 10, No 1. https://jedm.educationaldatamining.org/index.php/JEDM/article/view/158