Eswin2eng

Toolkit ESWin 2.0

for development of Expert Systems

This software is intended for creation and using of Expert Systems for solving of different tasks in diagnostics, planning, forecasting and so on.

Software is based on technology of Hybrid Expert Systems of company Insycom Ltd. This technology is characterized by:

Representation of knowledge by frames, rules and linguistic variables,

Backward fuzzy inference engine,

Possibility to execute external programs,

Import of facts from databases by SQL-query during inference,

For comment of questions and comment of results at solving of tasks the graphics files (gif, avi and html) may be used.

Toolkit consists of:

ESWin - Expert Shell for solving of tasks by interpretation of Knowledge Bases (for developers),

ESWinUs - Interpreter of Knowledge Bases for End Users,

EdKB, KlbEdit - Editors of Knowledge Bases,

KBView - program for viewing and diagnostics of Knowledge Bases.

The following methods of knowledge representation are used in ESWin:

rules with fuzzy logic and backward inference,

frames for describing of structure of area and dialogue with user,

linguistic variables for describing of fuzzy concepts in frames.

Facts for inference are got from user and Data Bases by SQL-query. Pictures (in formats GIF, AVI, HTML) may be used at inference as comments to questions and illustrations to results. The external programs may be used as sources of facts and for other goals. These programs may be executed as results of rules.

Components of ESWin

Developer of Expert System User
Example of Knowledge Base in ESWIn:

TITLE = ES for choose of knowledge representation method (part)

COMPANY= Computers Department of NSTU

FRAME = Goal

Knowledge representation method: ()

ENDF

FRAME = Type

Tasks [For what tasks ES will be used?]:

(diagnostics; configuring; planning; monitoring; forecasting)

ENDF

FRAME = Area

Used in [What area is planning?]: (medicine; computers; CAD/CAM; ; legacy)

ENDF

FRAME = Description of area

Number of concepts [How many concepts are in area?]:

(less then 10; from 10 to 100; from 100 to 1000; more then 1000)

Structures are needed: (yes; no)

Steps of solving: (1; 2; some)

ENDF

FRAME=Acting

Parent:

Program: ()

Message: ()

Frame: ()

Delete: ()

Execute: ()

Rule: ()

EndF

RULE 1

=( Description of area. Number of concepts; less then 10)

DO

=( Description of area. Structures are needed; no) 100

ENDR

RULE 2

=( Description of area. Number of concepts; more then 1000)

DO

=( Description of area. Structures are needed; yes) 100

ENDR

RULE 3

=( Description of area. Number of concepts; from 100 to 1000)

DO

=( Description of area.Structures are needed; yes) 90

ENDR

RULE 4

= (Area. Description of area; medicine)

= (Type.Tasks; diagnostics)

DO

=( Knowledge representation method; Frames) 100

=( Knowledge representation method; Fuzzy rules) 90

ENDR

RULE 4

= (Area. Description of area; computers)

= (Type.Tasks; configuring)

DO

= (Knowledge representation method; Frames) 100

= (Knowledge representation method; Fuzzy rules) 70

= (Knowledge representation method; Semantic nets) 70

MS(Acting.Message; probably the combination of offered

methods is needed)

ENDR

RULE 5

=( Description of area. Structures are needed; yes)

DO

= (Knowledge representation method; Frames) 100

ENDR

RULE 6

=( Description of area. Steps of solving; some)

DO

= (Knowledge representation method; rules) 100

ENDR

Demo version of ESWin 2.0 is available for download from http://www.insycom.ru

We are opened for cooperation in distribution and development of this product.