Learning from stories: indexing and reminding in a socratic case-based teaching system for elementary school biology
Daniel C. Edelson · 1993
Good teachers teach with stories. To reproduce this effective teaching technique in a computer-based learning environment, we have developed the case-based teaching architecture. A case-based teaching system uses artificial intelligence techniques adapted from case-based reasoning research to teach by presenting stories in context. That context is established by a task environment which provides a student with a naturally motivating task and an engaging environment for pursuing that task. A storyteller monitors the student's interactions with the task environment, looking for opportunities to present stories that will help the student to learn from his or her situation. Learning from stories provides students with cases that support the natural process of case-based reasoning. The challenge of constructing a case-based teaching system lies in developing a scheme for indexing the stories in the storyteller's library and in implementing algorithms that will allow the storyteller to identify appropriate stories when they are relevant. To retrieve these stories, a case-based teaching system employs reminding strategies. Reminding strategies enable a storyteller to identify and retrieve stories to serve specific goals. Reminding strategies rely on a sufficiently expressive indexing scheme for labeling the stories in a story library. In this dissertation, I present the case-based teaching architecture and propose several general reminding strategies. These have been used in the development of Creanimate, a case-based teaching system that teaches animal adaptation to elementary school students. Its task environment invites a student to create a new animal and engages the student in a discussion of how his or her invented animal might survive. Its storyteller possesses a library of dramatic video clips showing actual animals in the wild. These video stories are presented as illustrations of the principles that arise in the discussion of the student's animal. Creanimate uses thought-provoking questions called explanation questions as the basis for dialogues that establish a context for learning from stories. Creanimate employs three general reminding strategies: example reminding, similarity-based reminding, and expectation-violation reminding. Its story library is indexed according to the principles of animal adaptation that the stories illustrate.