[ home | resume | contact | science | art | personal | email ]

Davies, J., Goel, A. K., & Yaner, P. W. (submitted). Proteus: Visual analogy in problem solving. Submitted to Artificial Intelligence.

Cite this for: Publisher: (unpublished)

BibTex Entry:

Choose download format: PDF

From the Visual Analogy research theme.

Abstract

This work examines the hypothesis that visual knowledge alone is sufficient for analogical transfer of problem-solving procedures in some domains. It develops a computational theory of visual analogy in problem solving which has been implemented in a computer program called Proteus. Proteus provides two main things. Firstly, it provides a content account for visual analogy in problem solving, along with a corresponding vocabulary and data structures for representing the knowledge content. Secondly, Proteus provides a process account for visual analogy in problem solving, along with corresponding methods and algorithms. Proteus addresses all major subtasks of analogical problem solving. In addition, it identifies a new subtask in the task stucture of analogical problem solving: dynamic generation of new mappings between the intermediate knowledge states in the source and the target cases when a step in the transferred procedure creates a new object. Finally, by examining the limitations of use of visual knowledge alone, Proteus helps identify the functional roles of (non-visual) causal and functional knowledge in analogical problem solving. Back to Jim Davies's list of publications.

JimDavies ( jim@jimdavies.org )