LEARNING PROCEDURES BY SIMILARITY: MORE ADVENTURES WITH METAMOUSE
Date
1991-02-01
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
METAMOUSE is a system for inducing graphical procedures on the basis
of the examples it is shown. It is capable of inducing procedures
containing variables, conditional branches and loops. A procedure can
be induced from the first example shown, and can be generalized in
accordance with the examples the system encounters later on.
METAMOUSE starts predicting user actions as soon as it is possible,
thus facilitating the process of procedure induction and also
lessening the possibility of user errors. The process of procedure
induction is based on domain knowledge, and the learning algorithm can
be classified as an instance of learning by explanation.
The goal of this research is to investigate the effect of
similarity-based techniques application to procedure induction. Such
learning techniques are rather independent of the area of application.
Furthermore, machine learning techniques based on similarities in
examples have not been used in procedure induction before, and
therefore, it is to be expected that the results of this research will
be very stimulating.
Description
Keywords
Computer Science