[comp.society] ITS Knowledge Acquisition Info Needed

grassi@hpindda.HP.COM (Maurizio Grassi) (08/24/88)

[Note: ITS = Intelligent Tutoring System -- ed.]

My name is Maurizio Grassi, I am a student at the Polytechnic of Milan
in Italy.  I am developing a thesis about a prototype of an ITS system.

My main interest is about your methodology in knowledge acquisition.

In the project I am involved, I have to explore this problem and try to
create an implemented model of an interactive knowledge acquisition
tool.  I am working with two other students.  They solved the problem of
guiding the student in learning and how to recover from his errors, and
implemented a prototype of ITS which teaches two topics:

	- English verbs, 
	- semantic nets.

I have to generalize this work we design together and implement a tool
which creates the knowledge base for the system.  I would really
appreciate it if someone would send me some information about projects
or interesting references which can help me to design my work.

A report about the project follows.

Maurizio Grassi

- - - - - - - - - - - - -

THE "EDUCATION EXPERT ENVIRONMENT"

This research analyzes models of knowledge handled to provide an
intelligent computer aided instructor.  The methodology we are
developing tries to work out the principles of teaching/learning
processes on which pedagogic tools need to be based.  

These principles were formalized by Burton and Brown in:  " An
investigation of computer coaching for informal learning activities "
(see Sleemann and Brown, Intelligent Tutoring Systems, Academic Press
1982).  From this basis we want to build a shell-system which is a
global environment in instruction.

Teaching subjects can be divided in structured and non-structured, i.e.
knowledge can be achieved by students in an incremental step-by-step way
or in an unstructured manner.  Our work designs a tool which solves the
problem of structures learning.  The key points that characterize the
intelligent tutoring system we are modeling, are:

  - being able to answer a variety of user questions and
    explaining its arguments using a model of the application domain; 
  - giving advice in order to proceed with a task; 
  - selecting and ordering material and choosing the most appropriate 
    teaching strategy; 
  - having a student's model; 
  - optimizing the learning process.  

With the project, called Education Expert Environment, we want to
monitor on the one hand the teacher in producing the subject and on the
other the student in using this material.  In order to create a flexible
tool, we divide the problem into three parts:

  - the development of a course planning unit in which to acquire 
    the knowledge (IN DESIGN); 
  - the project of an inference engine about student's move (DONE); 
  - the study of a manager of student's learning modality (DONE); 

The knowledge is recorded in a genetic graph (see Goldstein, " The
genetic graph:  a representation for the evolution of procedural
knowledge ", International Journal of Man-Machine Studies 1979) we
organize in three levels easy to handle:

  - the "island-network" in which the limits of the argument are defined; 
  - the "island-model" in which the contents of the subjects are referenced;
  - the "concept-model" which is the specific learning unit.

The DESIGN UNIT 

  The teacher, using a knowledge acquisition tool, plans the course 
  and describes the teaching process.  The "concept-model" of a particular 
  argument is developed.  Each model, which plays the role of rule-based 
  domain expert, is embedded in the "island-model", the brick of the 
  "island-network" that defines the global courseware.
  
The TEACHING-LEARNING UNIT 

  The student, using the teaching/learning environment, is guided 
  through the "island-network", the learning contest, by an expert 
  manager which adapts its advice and explanation to the background 
  knowledge and experience of the user.

The INFERENCE MACHINE 

  Each concept is a framework to test student's knowledge.  An expert 
  "teacher", in a differential way, debugs student's errors and builds 
  a model of his learning activity.  This model is used by an "orator" 
  which defines the answers and examples to be proposed to increase 
  the learner's knowledge.  An "evaluator" judges the quality of each 
  move to investigate the student's capability.

- - - - - - - - - -