Information Technology Reference
In-Depth Information
•
Natural language is not used as such and a formal-based language is preferred, in the tradition of
languages such as KIF, that are thoroughly employed in artificial agents communication. These
formal languages prevent problems that rise from the ambiguity intrinsic to natural language.
•
When one of the agents is human, then his/her knowledge is opaque not only to his/her interlocu-
tor (here, the system) but also to the designer of the system. Therefore, the designer must build,
in his system, a series of 'guessing' strategies, that do not necessarily fathom the interlocutor's
state of mind, and might lead to failure in dialog. Whereas, when both agents are artificial, they
are both transparent to the designer, if not to each other. Thus, the designer embeds, in both, tools
for communication that are adapted to their knowledge level. The designer might check, at any
moment, the state variables of both agents, a thing he or she cannot do with a human.
These two restrictions tend to simplify the problem, and more, to stick to the real core of the task,
i.e. controlling acquisition through interaction.
the theoretical
fra Mework
Agents f rame
Our environment focuses on a situation where two cognitive artificial agents are present, and their sole
interaction is through dialog. During this relationship, an agent will play the role of a teacher and the
other will momentarily act as a student. We assume they will keep this status during the dialog ses-
sion. Nevertheless, role assignation is temporary because it depends on the task to achieve and on each
agent's skills. The 'teacher' agent must have the required skill to teach to the 'student' agent, i.e.
to
offer unknown and true knowledge, necessary for the 'student' to perform a given task
. Convention-
ally, 'student' and 'teacher' terms will be used to refer, respectively, to the agents acting as such. The
'teacher' aims at 'freely' offering a set of predetermined knowledge to the 'student'. This, naturally
subsumes that agents cooperate. Thereby, no erroneous data will be exchanged and agents will attempt,
using all means they can, to satisfy their interlocutor's expectancy. Nevertheless, as in a natural situa-
tion, the 'student' could be not really self-motivated and by this way making harder the 'teacher's task.
For instance, the 'student' could provide indefinite data to the 'teacher'.
Knowledge Base Properties
First-Order Logic.
Each agent owns a knowledge base (KB), structured in first-order logic, with func-
tions, so the knowledge unit is a
formula
.
First-order logic has been preferred to propositional logic,or description logic, because of the expressive
power of predicates, and the existence of functions was necessary to the nature of our first test corpus,
which was in physics (teaching laws of mechanics). However, functions have been abandoned because
of intrinsic difficulties, and we changed the corpus into a basic science corpus. Since quantifiers not
being tested, the traps related to them in first-order logic have been avoided. So, first-order logic here
mostly appears because FR modeling, introducing particular predicates (functional roles), has driven
us to use this level of expressivity.
Search WWH ::
Custom Search