You are on page 1of 3

VanLehn, K. (2006) The behavior of tutoring systems.

International Journal of A
rtificial Intelligence in Education. 16, 3, 227-265. (e-mail Sergey Sosnovsky at
"sosnovsky {AT} gmail {DOT} com" to obtain this paper)
o What is the purpose of the article?
a common standard , inner loops vs outer loops definition

o What do the following terms mean? Knowledge Component, Learning vs.


Physical Event, Tutoring Strategy, Inner vs. Outer Loop, Minimal Feedback vs. Er
ror-Correction Feedback vs. Hints.
Knowledge Component:
A task domain concept, principle, fact, etc. Any fragment of the persistent, dom
ain-specific
information that should be used to accomplish tasks.
Learning vs. Physical Event:
A mental event; the construction or application of a knowledge component, often
while trying to
achieve a task.

Inner vs Outer Loop:


Tasks are Outer loops - Steps in Tasks are inner loops

o What is the condition under which one can characterise the behaviour
of ITSs as consisting of an outer and an inner loop? How do the two relate to e
ach other, what are the responsibilities of each one, and which issues do they i
nvolve?
Inner loop
Minimal feedback on a step. In most cases, this means indicating only whether th
e step is
correct or incorrect.
Error-specific feedback on an incorrect step. This information is intended to he
lp the student
understand why a particular error is wrong and how to avoid making it again.
Hints on the next step.
Assessment of knowledge.
Review of the solution.
(1) selecting a task intelligently and (2) obtaining a rich
set of tasks to select from.
Outer Loop
The student selects the task from a menu of all tasks.
The tutor assigns tasks in a predetermined sequence.
Mastery learning: The tutor assigns tasks from a unit s pool of tasks until the st
udent has
mastered the knowledge taught by the unit.
Macroadaptive learning: The tutor tracks traits, including both unchanging trait
s such as
learning styles, and changing traits, such as correct and incorrect knowledge co
mponents. It
chooses a task based on the match between the task s traits and the student s traits
.

o Describe briefly the following systems:


+ Algebra Cognitive Tutors
+ Andes
+ AutoTutor
+ Sherlock
+ SQL-Tutor
+ Steve
Algebra Cognitive Tutor: Problem solving in a high-school algebra course.
Andes: Quantitative problem solving in introductory college physics.
AutoTutor: Several, including qualitative problem solving in college physics.
Sherlock: Troubleshooting a large piece of avionic electronic equipment.
SQL-Tutor: Constructing queries to relational databases in the language SQL.
Steve: Operating complex equipment, such as a room-sized air compressor.
anders- entry level physics -- > solve step by step, immediate f/b
auto tutor- advanced tutor, -ve f/b based system,, cmpre correct vs wrong and cr
eate f/b
sql tutor- teaches sql,student can select kind of f/b, hints can be selected by
students, assesment done by finegrain based on count of learn events.
steve- aveonics and ship, f/b will be generated wen der is damage to user or m/c
, at final it will describe hw to do, and wat we did. o/p pnts will be 1 to 100.
hints may give if needed.
-----------------------------------------

conventional - 2 sigma - least positive acc -


mastery - 1 sigma 84 percent - medium postve -
tutoring - average 98 percent - most positive acc -
initial aptitude test and interest in subject - basis for selection
compare initial and final - aptitude achievement correlation (acc)
2 sigma problem - due to
but 1 sigma possible - try to improve it
his students researched for possible methods to get better sigma.

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

* Corbett, K.R. Koedinger, and J.R. Anderson. Intelligent tutoring systems. In


M. Helaner and T.K. Landauer, editors, Handbook of Human-Computer Interaction,
Second Edition, pages 849 874. Amsterdam: Elsevier Science, 1997.(PDF)
o How does (Corbett et al, 1997) use the argument from (Bloom, 1984)?
Why?
Student should be more interactive with the tutoring.

o What is the research goal Corbett and his colleagues suggest and how
have they followed it themselves?
Tutoring tools should be for Training purpose not for AI purpose.

o What is the ACT-R Theory?


This theory assumes a fundamental distinction between declarative knowledge and
procedural knowledge.

o How does the analysis of ITSs in (Corbett et al, 1997) compare to th


e analysis in (VanLehn, 2006)? What is the architecture suggested?
o What is the instructional intervention at the:
+ Curriculum level?
+ Problem-solving support level?
The pedagogical module is responsible for structuring the
instructional interventions. This module may operate at two levels (Pirolli and
Greeno, 1998). At
the curriculum level it can sequence topics to ensure an appropriate prerequisit
e structure is
observed (Capell and Dannenberg, 1993) and individualize the amount of practice
at each level
to ensure the students master the material (Corbett and Anderson, 1995b). At the
problemsolving
support level, it can intervene to advise the student on problem solving activit
ies. Towne
and Munro (1992) outline five types of instructional interventions:
o What is the feedback like in the SHERLOCK system?
error feedback method

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

You might also like