Automatic Feedback and Algorithm Animation
Ari Korhonen, Lauri Malmi, Jussi Nikander and Petri Tenhunen



What I hear, I forget. 
What I see, I remember. 
What I do, I understand.  
	- Confucius

Currently the constructivistic learning paradigm is
widely accepted.  The key idea there is that the learner constructs
internal models of knowledge by himself.  An essential aid for
carrying out such a construction process successfully is getting
feedback when the model is applied to solve problems.

Providing feedback is laborious on large courses and therefore a
number of automatic feedback and assessment systems have been
developed.  In addition, it is evident that interactive working with 
a tool providing feedback aids the learning process also in small courses.

In this paper, we discuss different forms of automatic feedback and
how the feedback could be improved in context of learning style
models. In many of these models verbal and visual learners are separated. 
We, howevere, have divided automatic feedback into three categories:
verbal, visual, and operational. Methods should support different
learning models in teaching, and each of the categories should
be given equal attention while assessing the quality of automatic
feedback.

In general, verbal feedback includes all feedback represented in spoken
or written form.  In the simplest case it can be a grade or points
given from an exercise. Operational feedback guides the learning 
process and it can, for example, provide the student with additional 
learning material. It is a rather wide concept that includes all feedback 
that has a certain level of interaction with the user. Visual feedback 
refers to the part of feedback that can somehow be
visualized. For example, a model answer that demonstrates the steps of
solving an exercise in a visual environment. 

At Helsinki University of Technology, an automatic assessment system
for algorithm simulation problems has been used since 1991. 
In these problems the student simulates specified
algorithms by manipulating given data structures graphically.  The
result of his simulation is compared with the result of executing the
actual implemented algorithm on the data structures. Thereafter the
student is given feedback based on the comparison.  The old system is,
however, able to give only verbal feedback. Thus, a project to create
a new, more versatile system, called Matrix was launched in 1999. 
In this paper we describe the prototype for
automatic assessment system based on the new Matrix algorithm
animation framework.  The prototype combines verbal and dynamic
visual feedback in algorithm simulation exercises thus providing
considerably more information for the student.