|
|
Artificial Intelligence Problems
Division |
|
The Head of Division
Full Professor, Doctor of Sciences, Dimitry A. Pospelov
Our contact address
post: Artificial Intelligence Problems Division,
Computer Centre Academy of Sciences of Russia,
Vavilova 40,
GSP-1, 117967, Moscow, Russia
phone: 007 +095 135-6193 or 135-3298
fax: 007 +095 135-6159
e-mail: pospelov@ccas.ru
or
khor@ccas.ru
or
averkin@ccas.ru
Available topics:
- Theoretical background of Knowledge Based Systems
- Knowledge Representation in New Generation AI-systems
Special knowledge representation languages family,
based on frame & production rules from the one side
and using demonization C-procedures - from the
another. Special environment for edit, compiling
and debuging KB.
- Fuzzy logics for expert systems
The inference module of expert system uses family
of fuzzy logics. They are given by the axiomatic
system of triangular norms and depends on several
parameters. This fact allows to change the logic on
each step of inference and makes inference rather
flexible.
- Modern communication in AI-systems
Special models of discourse including user's models
and models of explanation, argumentation etc. based
on ATN-approach.
- Models & tools of Inference
Dialog CAD of inference engine including generation
of its executable code.
- Fuzzy inference simulation module
New methods are develloped for inference modules
designing for expert systems with high possibilities
of fuzzy logic inference and knowledge aquisition,
using probabilistic simulation. We must choose the
best logic between severel thousands fuzzy logics
for the given knowledge base and for the given
length of inference. For this purpose we use the
probabilistic simulation of fuzzy knowledge base and
of inference trees. Using the results of simulation
we can compare the result of inference for different
logics and choose the best logic for the real expert
system.
- Technology of Knowledge Based Systems Design & Implementation
- Knowledge Based Models of Life Cycle for AI-systems
Development of technological knowledge bases for
supporting, planing and managing AI-CAD based on cognitive
graphics approach.
- Fuzzy logics aquisition module
Expert system uses fuzzy logics, which simulates
the reasoning process of expert. So it should be
supplied by special methods of fuzzy knowledge and
fuzzy logics aquisition. Special cognitive graphic
interface is used to show to the expert the truth
tables of fuzzy (multi-valued) logics, the
distibution of truth values in knowledge base etc.
Expert can use special graphic editor to change the
logic and to see the possible results of inference
before using the logic in the expert system. This
module is still unknown in expert system shells and
it is called "fuzzy logics aquisition module".
- Cognitive Graphics in Knowledge Based Systems Design &
Implementation
Special knowledge bases supporting processes of create
and edit graphical patterns describing main
components of AI-systems based on Chernoff's faces.
- AI-software
- Design & Implementation of REFAL - Rewriting Rules Based
Programming System
REFAL is based on Markov's approach to recursive
functions. Sourse language have flexible
posibilities for list processing and may be expanded
by external C-functions. REFAL used as metalanguage
for knowledge representation languages
implementation, in symbolic processing etc.
- Design & Implementation of PILOT/2 - Knowledge Representation
Tool for second generation expert systems
PILOT/2 is the language based on production rules
and frames. The OPS5 and ART are It's prototypes.
PILOT-compiler generates C-program for Windows
environment.
- Design & Implementation of ATNL - communication models
description language
ATNL is based on Woods' Augmented Transition
Network model. ATNL-compiler is implemented on REFAL
and generates REFAL -program as a target program.
This language is used for natural language
communication, explanation, argumentation etc.
- Design & Implementation of FrameBrush - Cognitive Graphics
Editor
Special cognitive graphical editor FrameBrush
produces frame representation of graphical patterns.
- Design & Implementation of PiES WorkBench - Intelligent
CASE for Knowledge Based Systems
PiES WorkBench - instrumental knowledge based
system oriented on Expert Systems design and
implementation. PiES WorkBench is in progress now
and Windows is the basic environment here. All
managment, technological and special knowledge in
this system represented in frame-oriented DBMS
package FRAME/2 and KRL PILOT/2. Cognitive graphic
approach is widly used in PiES WorkBench at the
managment, design and implementation stages of
prototyping applied expert systems.
- Fuzzy expert system shell
The prototype of the shell is designed. The new
elements for the module are flexible family of fuzzy
logics, fuzzy logics aquisition module and fuzzy
inference simulation module implemented in C++ and
in Windows 3.1 environment.
- Applied Knowledge-Based Systems
- Applied fuzzy expert systems in medicine and psychology
The prototype of ES in neurology (1500 rules) is
made in Russian Centre of reabilitation for
neurologic deceases. We are working together with
Phycology faculty of Moscow State University to
create psychological diagnostic system, using the
fuzzy knowledge acquisition methods. For this
purpose we use fuzzy expert system shell.
- Expert Systems for Personal Psychological Diagnostics
Applied ES "Cattell" produces an individual textual
person's portrete based on test of Cattell.
Applied ES "Ayzenk" produces an individual textual
person's portrete based on modification of Ayzenk test.
Applied ES "Lusher" produces an individual textual
person's portrete based on color test of Lusher.
- Knowledge Acquisition Psychological Support (KAPS) Tool Kit
KAPS includes ESs "Cattell" and "Lusher" now,
expert system "Ayzenk" is in progress. All these systems
will be included in KAPS Tool Kit that based upon
our original approach to knowledge acquisition
process.
- Intelligent Computer-Aided Instruction Systems based on
Socratic's conversations
WHY is a prototype of these systems.
We are interested in Your opinion in these domains,
open to contacts and hope for fruitful results of collaboration for both
sides. Communicate with us NOW!!!
Copyright © 1996, Khoroshevsky
