e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.

htm

[Home: Demos and Tutorials][e2gRuleEngine Mini-Course Index]

Building and Using Expert Systems: a Mini-Course Introducing the e2gRuleEngine/e2gDroid Expert System
Shell

e2gRuleWriter Tutorial and Reference Documentation: Building, Validating and Maintaining e2gRuleEngine Knowledge Bases
in a Decision Table Format

e2gRuleWriter Tutorial and Reference Table of Contents

Overview of the e2gRuleWriter tutorial

Introduction to decision tables
Decision table definition
Early uses
Decision tables and rule-based expert systems
Limited, extended and mixed formats
Generating all possible condition stub combinations
Decision table simplification
Rule validation
Rule aggregation during conversion to knowledge base format
Benefits and limitations of decision tables for developing knowledge bases
Acquiring and using the e2gRuleWriter software
e2gRuleWriter quick start
The development process
e2gRuleWriter navigation
The "weather" example problem
Defining conditions
Defining actions
Defining rules
Generating the knowledge base
Rule generation and rule simplification
Detailed tour of e2gRuleWriter features: condition/action modes
Condition/action row controls
Editing row controls
Condition/action list value controls
Default value controls
Condition prompt controls
Decision table load/save controls
Detailed tour of e2gRuleWriter features: rule mode - rule construction
Rule column controls
Editing rule controls
Automatic condition stub generation
Decision table load/save controls
Detailed tour of e2gRuleWriter features: rule mode - simplifying and validating rules
Simplification/validation option check boxes
Simplification/validation execution
Detailed tour of e2gRuleWriter features: rule mode - generating an e2gRuleEngine knowledge base
Knowledge base options
Knowledge base generator execution
e2gRuleWriter examples
Limited format example (yesno.htm and yesno.kbt)
Auto diagnosis example (auto.htm and auto.kbt)
Auto diagnosis example - French translation (autofr.htm, autofr.kbt and french16.txt)
Weather example - simplified Chinese translation (weatherch.htm, weatherch.kbt and chinese16.kbt)
Wine choice example (wine.htm and wine.kbt)
e2gRuleWriter operating system and user interface considerations

Overview of the e2gRuleWriter tutorial

The objectives of this tutorial are to:

1. Introduce classic decision tables as a well established systems analysis support tool.
2. Present the benefits and limitations of using decision tables to develop and maintain expert system knowledge bases.
3. Introduce the e2gRuleWriter software as a tool for building and validating decision tables that generate e2gRuleEngine knowledge bases.
4. Provide a variety of examples that demonstrate the capabilities of e2gRuleWriter.
5. Provide detailed reference documentation for e2gRuleWriter.

Introduction to decision tables

Decision table definition: A decision table is a rectangular array of cells with rows representing conditions and actions and columns representing rules. The
condition part of the decision table is called the condition stub shown in yellow in the figure below and the action part the action stub shown in green. Each column
represents a rule, and when all of the conditions in a column are satisfied, the action or actions designated in that column represent the recommended decision.

1 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Rule1 Rule2 Rule3

Condition1
Condition2

Action1
Action2

Early uses: Representing decision logic with decision tables predates the advent of expert systems, and probably the invention of the computer. They have long
been employed in printed form as a device for communicating decision logic that is both transparent and unambiguous. They have also been used as a systems
analysis tool for documenting and verifying software logic, especially in business applications, and their use for that purpose is described in many systems analysis
textbooks. There is some variation in both terminology and technique in these texts. In this tutorial we will generally follow the treatment provided in the following
reference:

Analysis and Design of Information Systems, 2nd ed.

James A. Senn
© 1989 by McGraw-Hill, Inc.
ISBN 0-07-056236-9

This book is no longer in print, but is available in libraries and in the used book market, and is highly recommended. Many systems analysis and design texts cover
the topic, but most provide only cursory coverage compared to Senn.

The high cost of developing software has long motivated an interest in automated tools to support the computer programming process. Many of the attempts at
automatic program generation, developed as early as the 1960's, employed decision tables as an input to software that could generate parts of computer programs.
Rule columns in a decision table can easily be converted into if...then... program statements, and the decision table itself is an effective tool for defining and verifying
decision logic with non-programmers. Decision tables are often employed in modern business rules applications, either as an input to program generators or to
provide the input to a rule engine that operates as a forward-chaining inference engine.

Decision tables and rule-based expert systems: The connection between decision tables and rules in a rule-based expert system is easily seen in the following
decision table adapted from the eXpertise2Go Mini-Course Knowledge Engineering module. "Y" and "N" in the condition stub represent "Yes" and "No" ("T" and "F"
could be used representing "True" and "False") and an "X" in the action stub means the action is recommended. "-" in a cell means the condition or action is
irrelevant:

Rule1 Rule2 Rule3

does the starter run? Y Y N
do you smell gas when the starter cranks? Y N -

replace the dead battery - - X

refuel the car - X -
wait 10 minutes, restart flooded car X - -

Limited, extended and mixed formats: This form of a decision table is called a Limited Entry Form, employing only Y, N, X, and blank or irrelevant entries. An
Extended Entry Form of the table allows multiple values for conditions or actions. Here is the automobile diagnosis decision table in this alternative format:

Rule1 Rule2 Rule3

the result of trying the
the car cranks normally the car cranks normally nothing happens
starter
not present while trying the
the smell of gasoline present while trying the starter -
starter
wait 10 minutes, restart flooded recharge or replace the
the recommended action refuel the car
car battery

Although either decision table form is easily mapped into expert system rules, the extended entry form is probably most effective for this purpose as the following
rules extracted from the three columns of the table illustrate:

Rule1 Rule2 Rule3

If the result of trying the starter = the car cranks If the result of trying the starter = the car cranks If the result of trying the starter = nothing happens
normally and normally and Then the recommended action = recharge or replace
the smell of gasoline = present while trying the starter the smell of gasoline = not present while trying the the battery
Then the recommended action = wait 10 minutes, starter
restart flooded car Then the recommended action = refuel the car

It is possible to combine the limited and extended entry formats in a Mixed Form decision table, and e2gRuleWriter supports this approach.

Generating all possible condition stub combinations: The number of possible decision table rules that can be generated from a set of conditions is the product
of the number of values of each of the conditions. To make sure all possible rules are considered, a step-by-step procedure for constructing condition stubs is useful.
Here is an example that illustrates one such procedure:

Condition1 with 2 values: T and F for True and False

Condition2 with 3 values: A, B and C
Condition3 with 2 values: 5 and 10

There are 2 x 3 x 2 = 12 possible combinations of these three conditions' values. To generate the condition stub:

1. Write down the product of value counts, dropping the count for the first condition and adding a one (1) after the last condition: 3 x 2 x 1 for this example.
2. To produce the first row of decision table values, write down each value of Condition1, repeating the value according to the count from the product of the
value counts: 3 x 2 x 1 = 6, so each value will be repeated 6 times: T T T T T T F F F F F F and there will be 12 columns.
3. Drop the left most value from the product of value counts and write down each value of the next condition, repeating the value according to the count from the
remaining value counts: 2 x 1 = 2 so each value will be repeated 2 times: A A B B C C This sequence is repeated until all columns (12 in this case) of the
table are filled in. The second row of the table will have the following appearance: A A B B C C A A B B C C again including 12 columns.
4. The process is repeated until all conditions have been incorporated into the table. Dropping the next value from the product of the value counts leaves a 1, so
the final condition values are written down once, with the sequence repeated until the 12 columns of the table are filled in: 5 10 5 10 5 10 5 10 5 10 5 10.

2 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Inserting these sequences of values for the three conditions into a decision table's condition stub produces the following result:

Rule 1 Rule 2 Rule 3 Rule 4 Rule 5 Rule 6 Rule 7 Rule 8 Rule 9 Rule 10 Rule 11 Rule 12
Condition1 T T T T T T F F F F F F
Condition2 A A B B C C A A B B C C
Condition3 5 10 5 10 5 10 5 10 5 10 5 10

At this point, action rows can be added to the table and each rule column examined to determine the appropriate action value(s). If a rule column is considered
irrelevant it is dropped from the table.

Decision table simplification: It will often be possible to simplify a decision table, eliminating unnecessary conditions or combining rule columns before generating a
knowledge base. The table simplifications supported by the e2gRuleWriter will be presented in more detail in a later section of this tutorial, but to illustrate the point,
consider the following example table:

Rule1 Rule2 Rule3

Condition1 Y Y N
Condition2 Y Y Y
Condition3 Y N Y

Action1 X X -
Action2 - - X

Rule1 and Rule2 have the same actions, have the same condition values for all but one condition (condition3) and all values of condition3 are represented by the two
rules. Condition3 is thus irrelevant, and can be set to blank ("-") in Rule1 and Rule2. Once this is done, the two rules are identical and either one of the rules may be
removed from the table.

Rule validation: It is also possible to examine rules in the decision table for errors and contradictions. In the following table:

Rule1 Rule2 Rule3

Condition1 Y N Y
Condition2 N Y N

Action1 X X -
Action2 - - X

Rule1 and Rule3 have identical conditions, but different actions -- a contradiction if Action1 and Action2 are mutually exclusive. The possible contradiction should be
resolved before generating a knowledge base from the decision table. If it is possible for both actions to be taken simultaneously, the table should be simplified by
combining the two rule columns:

Rule1 Rule2
Condition1 Y N
Condition2 N Y

Action1 X X
Action2 X -

Rule aggregation during conversion to knowledge base format: Depending on the format permitted for rules used in a specific expert system shell, it may be
possible to aggregate decision table rule columns when the knowledge base is generated. Since all conditions in a decision table column must be satisfied for the
action or actions in that column to be recommended, the conditions are connected by the "and" logical operator. Expert system rules typically allow use of the "or"
logical operator:

Rule1 Rule2 Rule3

Condition1 Y - -
Condition2 - Y -
Condition2 - - Y

Action1 X X X
Action2 - - -

In this table, each rule has a single condition, and in all three rules the same action is concluded. Therefore, the three rules are equivalent to the following single rule
that could be generated for an expert system knowledge base:

IF Condition1=true OR
Condition2=true OR
Condition3=true
THEN Action1=true

Benefits and limitations of decision tables for developing knowledge bases:

What's good:

The decision table presentation is simple to understand because it is transparent and unambiguous. Each column may be interpreted separately, and has a
single interpretation: if the conditions are satisfied, the specified action or actions should be taken.
A decision table can be interpreted by individuals who have knowledge of the problem domain, but have little or no background in systems analysis or
software development.

3 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Knowledge base maintenance may be easier to accomplish without errors if the knowledge base is derived from a decision table. Common errors in
knowledge base maintenance such as editing that creates inconsistent spelling of attribute names or values may be easy to detect or prevent with
well-designed decision table software.

What's not so good:

Expert system rules typically provide a higher level of abstraction than decision table rule columns: the ability to use "or" logical operators illustrated in the last
section provides one example. It may take many columns in a decision table to represent a single expert system rule. Extensions to the classic decision table
format, permitting "or" logical operators, "else" actions and other features are sometimes implemented to minimize these limitations.
Using a decision table to produce a knowledge base requires learning another piece of software beyond the expert system shell.
The conversion from the decision table to the knowledge base is typically one-way: you can't derive the decision table from the knowledge base. If the
knowledge base is edited outside the decision table software (perhaps to employ inference engine capabilities not accessible from the decision table format),
many of the knowledge base maintenance advantages are lost.

Acquiring and using the e2gRuleEngine/e2gRuleWriter software from eXpertise2Go.com

The conditions of use for e2gRuleWriter are similar to the e2gRuleEngine conditions of use:

Conditions of use that you accept when you download e2gRuleEngine/e2gRuleWriter from eXpertise2Go.com

Purpose of e2gRuleWriter and disclaimers: This software has been developed and is being distributed for educational and experimental purposes. It is a work
in progress, and will continue to be a development prototype for some time to come. No claims are made for its serviceability, and eXpertise2Go.com assumes
no responsibility for financial or other harm that occurs due to the use of e2gRuleWriter or knowledge bases developed with the software to work with
e2gRuleEngine or due to decisions made based on use of these knowledge bases. While reasonable effort is being expended in the ongoing effort to test and
debug the program, when you download and use eXpertise2Go software you become an informal member of the test team!
You may not sell e2gRuleWriter: e2gRuleWriter is free software: eXpertise2Go.com isn’t selling it, so you can’t either. It is OK to use e2gRuleWriter to
generate knowledge bases you develop for compensation, but your client must understand that e2gRuleWriter is not part of what they are paying for.
Re-engineering/reverse engineering e2gRuleWriter is prohibited: You agree that you will not de-compile and/or modify the e2gRuleWriter application.
Use of e2gRuleWriter to develop commercial Web sites is allowed: You may use e2gRuleWriter to develop e2gRuleEngine knowledge bases that are
incorporated in commercial Web sites.

The download package: For ease of installation, the download is in the form of a compressed (.zip) file. This file contains the software for both e2gRuleWriter and
e2gRuleEngine in the form of archives (.jar files) and a collection of decision table and .htm files that will be useful in testing your installation and learning to use the
software. Download e2g3g.zip by clicking on the link below the list of files in the .zip and choosing the "save file" option. Then, unpack e2g3g.zip into a subdirectory
using a program like WinZip. In some versions of Windows, you can unpack the files by right-clicking the e2g3g.zip file name, then clicking on the "extract all files"
option. You may also download the files individually by right-clicking the name of the file you want in the list below, then clicking the "save link" option and specifying
the subdirectory into which the file should be saved. When you examine your target subdirectory, you will find e2gRuleEngine.jar and e2gRuleWriter.jar included with
the .htm, .kb and .kbt files. Do NOT unpack these archives: they are intended to be used as .jar files, and none of the test files will work if they are unpacked. Here
are the files that should be extracted into your subdirectory:

File Description
e2gRuleWriter.jar e2gRuleWriter software: keep the .jar, do not extract!
e2gRuleEngine.jar e2gRuleEngine applet software: keep the .jar, do not extract!
test.htm generic .htm file with e2gRuleEngine applet tags
auto.htm .htm file with applet tags to run auto.kb
auto.kbt decision table to generate auto.kb
auto.kb knowledge base
autofr.htm .htm file with applet tags to run autofr.kb (Unicode)
autofr.kbt decision table to generate autofr.kb
autofr.kb knowledge base (Unicode)
french16.txt translation table to incorporate in autofr.kb (Unicode)
weatherch.htm .htm file with applet tags to run weatherch.kb (Unicode)
weatherch.kbt decision table to generate weatherch.kb
weatherch.kb knowledge base (Unicode)
chinese16.txt translation table to incorporate in weatherch.kb (Unicode)
wine.htm .htm file with applet tags to run wine.kb
wine.kbt decision table to generate wine.kb
wine.kb knowledge base
yesno.htm .htm file with applet tags to run yesno.kb
yesno.kbt decision table to generate yesno.kb
yesno.kb knowledge base

I agree to the eXpertise2Go.com conditions of use, download the .zip file

To use e2gRuleWriter (or e2gRuleEngine) you must have a recent version (1.6.x is current as this is written) of the Java Runtime Engine (JRE) installed and properly
configured on your computer. Installing Java to work with a Web browser will normally accomplish the proper installation and configuration including setting the
CLASSPATH variable.

e2gRuleWriter is a Java application, not an Applet. This means it does not run within a Web browser. To start the software, you should navigate to the subdirectory
that contains e2gRuleWriter.jar. Then, using a command line entry such as the Windows "Run" command or the MSDOS window, type:

java -jar e2gRuleWriter.jar

This should open the decision table software.

If you will be using e2gRuleWriter frequently on the same Windows computer, you may want to create a desktop shortcut. To do this, navigate to your subdirectory
that contains e2gRuleWriter.jar and right-click on the .jar file. Then, click on the Create Shortcut option. This will create a new entry in your subdirectory named
e2gRuleWriter.jar - Shortcut. Right click on this entry and select Open With and Choose Default Program... options. Then, select the Java Runtime Engine (Java
(TM) Platform SE Binary) as the program for this purpose and click the Always use the selected program to open this kind of file check box. Finally, drag the

4 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

entry on to the desktop and you will be able to double click the icon to start e2gRuleWriter.

e2gRuleWriter quick start

The development process: An e2gRuleEngine knowledge base is constructed and maintained with e2gRuleWriter through iterative application of the following
steps:

Define the conditions using e2gRuleWriter's condition mode options

Define the actions using e2gRuleWriter's action mode options
Define the rules using e2gRuleWriter's rule mode options
Simplify and validate the rules with rule mode options
Generate and examine the e2gRuleEngine knowledge base corresponding to the decision table
Generate and save the e2gRuleEngine knowledge base corresponding to the decision table
Test the e2gRuleEngine knowledge base using a Web page (.htm file) containing appropriate Java Applet tags

To provide a quick tour of the major features of the software, we will use e2gRuleWriter to redevelop the very simple "weather" knowledge base introduced in an
earlier eXpertise2Go mini-course module. This presentation assumes the reader is familiar with the operation of e2gRuleEngine and the format of an e2gRuleEngine
knowledge base.

Start up the software (java -jar e2gRuleWriter.jar) and you should see the following screen that represents the condition mode:

e2gRuleWriter navigation: Input screens representing three modes (condition, action and rule) are color coded: yellow for condition, green for action and blue for
rule. e2gRuleWriter starts operation in condition mode. Clicking the left mouse button on an action row (green row) changes to action mode. Clicking on a rule
number at the top of the table, or the gray row under the rule numbers (where rule names will appear) moves to rule mode. When you are in rule mode, clicking on a
row in the left-most column (under the Condition or Action headers) moves back to condition or action mode.

The program always initializes with an empty decision table containing 5 conditions, 3 actions and 5 rules. Conditions, actions and rules may be added or deleted
throughout the development process.

The currently selected row in condition or action mode is highlighted in blue. The currently selected rule column in rule mode is also highlighted in blue. A row or
column is selected by clicking anywhere in the row or column. It is possible to drag-and-drop rule columns to rearrange them: click on a rule number and drag the
column to the desired position while holding down the mouse button. Rules are output to the knowledge base in left-to-right order of the columns in the decision
table.

To change the displayed width of a rule column, click on the bar between rule number headers and, while holding down the left mouse button, drag the bar left or
right to decrease or increase the column's width.

Tooltips or "hints" are available for many of the data input fields. If the mouse pointer hovers above a control or data entry field, and a tooltip is available, it will be
displayed in the vicinity of the mouse pointer location. To disable tooltips, uncheck the "Tooltips enabled?" checkbox.

The "weather" example problem: To help clarify the e2gRuleWriter development process, let's first examine the demonstration e2gRuleEngine knowledge base we
will derive with our decision table example:

REM Test Knowledge Base (weather.kb)

RULE [Is it going to rain?]

If [precipitation] = "expected" and
[the expected temperature] > 32
Then [the recommendation] = "wear a raincoat"

RULE [Is it going to snow?]

5 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

If [precipitation] = "expected" and

[the expected temperature] < 33
Then [the recommendation] = "wear your boots" @ 95

PROMPT [precipitation] MultChoice CF

"According to the weather forecast, precipitation is:"
"expected"
"not expected"

PROMPT [the expected temperature] Numeric CF

"The expected daytime temperature (degrees Fahrenheit) is:"
"-30"
"120"

DEFAULT [the recommendation] = "wear whatever you want" @ 90

GOAL [the recommendation]

The elements of this example, in the terminology of a decision table are:

Conditions and possible condition values:

precipitation
expected
not expected
the expected temperature
number between -30 and 120

Actions and possible action values:

the recommendation
wear a raincoat
wear your boots
wear whatever you want (default value if the recommendation cannot be determined)

Information requested from the user (prompts) and attributes of the prompts

precipitation
multiple choice input
allow input of certainty factor (CF)
the expected temperature
numeric input
acceptable range of input: -30 to 120
allow input of certainty factor

Defining conditions: The appearance of the condition mode data entry screen after entering data describing the first condition is shown on the following screen:

The steps required to define the "precipitation" condition include:

1. With the first condition row selected, click in the text entry field labeled EDITING
2. Type the value precipitation
3. Press the ENTER key: you should see precipitation displayed as the first condition label in the decision table

6 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

4. Click in the text entry field labeled NEW LIST VALUE

5. Type the value expected and press the ENTER key: you should see expected displayed in the list box to the left of NEW LIST VALUE
6. Repeat, typing not expected in the text entry field labeled NEW LIST VALUE then pressing the ENTER key
7. Select MultChoice as the prompt type by clicking the radio button beside this value
8. Type the prompt stem: According to the weather forecast, precipitation is: in the text entry field labeled Prompt:, press ENTER

You may save your work at any time by clicking the Save decision table button. The dialog box below will appear, and you may navigate to the desired directory,
type the file name (which must have a .kbt file extension) and click the Save decision table button to complete the action.

Once you have saved the file, you can re-save by double clicking the file name in this dialog box. The dialog for loading a decision table (requested with the Load
decision table button) is similar in appearance. When a decision table is loaded from a .kbt file, the file name is displayed in parentheses at the end of the frame
header.

The decision table is saved in a format meaningful only to e2gRuleWriter. Attempting to modify this file in any way will likely render it ususable.

Similar steps are used to define the expected temperature as shown in the next screen image. Because this is a numeric quantity (determined by the selection of
the Numeric prompt type) you may enter minimum and maximum acceptable values beside the Range, ENTER: label. The e2gRuleEngine inference engine will not
accept a numeric input from the prompt unless it is in this range.

Also note the use of inequality operators in the list values: < 32 and > 33. When the relational operator in rules generated for e2gRuleEngine is the = for equality, the
operator is not explicitly entered with the list values. When any other relational operator is employed, it is entered with the list value.

Defining actions: Inputs that are accepted when defining conditions but not used by actions are disabled or "grayed" on the action mode screen:

7 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

The steps used to define "the recommended action" action as shown above include:

1.With the first action row selected, click in the text entry field labeled EDITING
2.Type the value the recommendation
3.Press the ENTER key: you should see the recommendation displayed as the first action label in the decision table
4.This condition will be the goal for the expert system consultation, so click the checkbox labeled Goal?
5.Click in the text entry field labeled NEW LIST VALUE
6.Type the value wear a raincoat and press the ENTER key: you should see wear a raincoat displayed in the list box to the left of NEW LIST VALUE
7.Repeat, typing wear your boots in the text entry field labeled NEW LIST VALUE then pressing the ENTER key
8.Prompts are entered with conditions, not with actions. If it is necessary to set the data type to "True/False" or "Numeric" rather than the default "Text" for an
action, this can be done with the radio buttons at the right side of the EDITING row of controls -- not necessary for this action
9. To specify a DEFAULT value for the recommendation provided as the result if e2gRuleEngine cannot determine a result from prompts or rules, type wear
whatever you want in the text entry field labeled DEFAULT VALUE, then press the ENTER key: you should see wear whatever you want displayed in the
DEFAULT field (the text field that has a gray background)
10. To assert this DEFAULT value with a Certainty Factor (CF) of 90%, choose 90 from the combo box at the left of the Set CF for DEFAULT button, then click
the button and @ 90 will be displayed at the end of the DEFAULT value
11. The DEFAULT CF value may be changed by selecting a new value in the combo box and clicking the set CF for DEFAULT button: if the selected CF is the
default 100%, no value is displayed beside the DEFAULT.

Defining rules: The final knowledge base construction step involves entering condition and action values to form the rules, then generating the e2gRuleEngine
knowledge base.

8 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

The steps used to build the rules shown above are:

1. Click on any rule number to navigate to e2gRuleWriter's rule mode

2. Click on the Rule 1 heading to select that rule for editing
3. Type the rule name Is it going to rain? in the text entry field named EDITING, then press the ENTER key: the rule name should appear below the rule
number in the decision table
4. Double click in the precipitation row of Rule1 to open a combo box containing all possible values of precipitation
5. Click the desired value: expected in this case
6. Double click in the temperature and the recommendation rows and select the values > 32 and wear a raincoat
7. Build Rule 2 in a similar fashion using Is it going to snow? as the rule title and selecting the precipitation, temperature and the recommendation values
shown in the screen above
8. To set the Confidence Factor (CF) of 95% for the recommendation in Rule 2, select the recommendation in the decision table, select 95 in the combo box
to the left of the button labeled Set CF for selected action, then click the button: the @ 95 value will be added to the action in the decision table
9. The CF for an action may be changed at any time by selecting the action, selecting a new value from the CF combo box then clicking the button: if the
selected value is 100%, the CF value will be blanked in the action (@100 is assumed for all actions by default)

Generating the knowledge base: The final step is to select the minimum Certainty Factor (CF) for the e2gRuleEngine inference engine to consider a value to be
known from the combo box labeled MINCF, then click the Display knowledge base button. A listing of the knowledge base will appear in a new window:

9 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

After reviewing the knowledge base, you may generate the same text as an e2gRuleEngine .kb file by clicking the Save knowledge base button. A file dialog will
open to allow selection of a knowledge base file name. This name must have a .kb file extension.

To test the weather.kb knowledge base, you will need a Web page with appropriate Applet tags. Here is a sample:

<HTML></TITLE></HEAD>
<BODY>
<APPLET CODE="e2gRuleEngine.class" ARCHIVE="e2gRuleEngine.jar" WIDTH=450 HEIGHT=300>
<PARAM NAME="KBURL" VALUE="weather.kb">
<PARAM NAME="DEBUG" VALUE="false">
<PARAM NAME="APPTITLE" VALUE="Weather Demonstration">
<PARAM NAME="APPSUBTITLE" VALUE="by The Weather Wiz">
<PARAM NAME="TITLECOLOR" VALUE="#FF0000">
<PARAM NAME="PROMPTCOLOR" VALUE="#FF0000">
<PARAM NAME="BGCOLOR" VALUE="#00FFFF">
<PARAM NAME="STARTBUTTON" VALUE="Recommend what to wear">
Java is required to view this page!
</APPLET>
</BODY>
</HTML>

If this .htm file is in the same subdirectory as the weather.kb file generated by e2gRuleWriter and the e2gRuleEngine.jar file, loading it into a Web browser will run
the weather expert system demonstration. When developing a knowledge base with e2gRuleWriter, an efficient procedure is to create a test .htm file for the
application and store it in a working subdirectory along with e2gRuleWriter.jar and e2gRuleEngine.jar. Generating the .kb file in this subdirectory lets you quickly
move between developing and testing your expert system application.

Rule generation and rule simplification: e2gRuleWriter supports automated creation of all possible condition stubs for a user-selected set of conditions. To
illustrate this capability with the weather example, hold down the ctrl key while clicking the precipitation and the expected temperature entries in the Automatic
rule generation list box to select both entries. Then, click the Generate rules for selected conditions button. There are two condition values for each of the
selected conditions, so four rules are possible including the two already defined. The following screen shows the result of automated rule generation:

10 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Rule names and actions could be added to the two new rules to extend the weather example.

Rule generation must be used with caution: the number of rules generated is the product of the number of values of each of the selected conditions, and it is possible
to generate a very large number of rules with this capability. e2gRuleEngine will display a dialog box showing the number of rules that will be generated and provide
an option to continue or cancel rule generation. If more than 500 rules would be generated, an error dialog is generated: clicking the "Cancel" key cancels the
proposed rule generation process. When developing a knowledge base it is easy to ignore combinations of condition values that could represent important rules, and
this function helps assure that all the possibilities are considered.

Finally, it is possible to clean up and validate the knowledge base by selecting options from the Rule simplification/validation... panel. To observe a simple
example, check the Remove unused condition/actions and incomplete rules checkbox, then click the Simplify and error check table button. You will be offered
a chance to save the decision table before the process continues. All unused rows (conditions and actions) and incomplete rules (including the two new rules
generated with the Generate rules... function) will be deleted, and a new window will open that summarizes modifications made to the decision table.

Detailed tour of e2gRuleWriter features: condition/action modes

This section presents a detailed description of the e2gRuleWriter's condition and action mode controls. When in the condition or action modes, the control
background color matches the background color of the condition or action rows in the decision table.

Conditions or actions are selected by left clicking anywhere in the desired row or by selecting a row, then holding down the left mouse button and dragging the
selection to a different row. The current properties associated with a condition or action will be displayed when its row is selected.

Controls that are not appropriate for use in the current state of the decision table are disabled and "grayed" (presented in a light gray color).

11 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Condition/action row controls:

1. Add a condition button: Adds a new condition row to the decision table. If a condition row is currently selected, the row is added before the selected
row. If a condition row is not selected (because an action row or one of the title rows is selected) the new row is added after the last condition row. A
vertical scroll bar is added to the decision table when the height of the table exceeds the display area.
2. Add an action button: Adds a new action row to the decision table. If an action row is currently selected, the row is added before the selected row. If
an action row is not selected, the new row is added after the last action row.
3. Move selected row up button: When e2gRuleWriter generates a knowledge base for e2gRuleEngine, rule conditions (the rule premise) and rule
actions (the rule consequent) are output in the order they appear in the decision table (top to bottom). To change this order, condition or action rows
may be repositioned by selecting the row, then clicking the Move selected row up button. Each click of the button moves the selected condition or
action up one row in the decision table. When the row is at the top of its section of the table (condition or action) clicking the button moves the row to
the bottom of its section. The e2gRuleEngine goal seeking process begins with the last goal in the GOAL statements in the knowledge base, then
works backward through the goals. GOAL statements are output to the knowledge base in Condition statement order (top to bottom) so rearranging
action rows that have the Goal? check box enabled allows control over this aspect of the inferencing process.
4. Delete selected condition/action button: Clicking this button deletes the currently selected condition or action. If any values of the condition/action
have been selected in rules, the following option is provided before the delete takes place:

Clicking OK in response to the warning deletes the row, cancel ignores the deletion request.
5. Tooltips enabled? check box: When this box is selected, a tooltip or hint will be displayed when the mouse pointer hovers near a control for which a
tooltip is available. Checking or clearing the tooltip check box in condition/action mode also checks or clears the box for rule mode.
Editing row controls:
1. EDITING text field: Typing a name in this field, then pressing the Enter key, assigns this name to the currently selected condition or action row. Text
may also be copied from another document and pasted into this field: use Ctrl-V to accomplish the paste action, then press Enter. The field is disabled
if no condition or action is selected. Once a condition or action has a name assigned, the name cannot be changed to blanks: the row must be deleted
with the Delete selected condition/action button if it is no longer needed. Only one condition and one action may have the same name: a row name is
ignored, and the EDITING text field cleared, if the name already exists as a row in the same stub (condition or action) of the decision table.
2. MAXVALS combo box: By default e2gRuleEngine allows only one value to be determined for a condition or action. Selecting a larger value from the
combo box generates a MAXVALS statement for this condition or action when the knowledge base is produced.
3. Text t/f Num radio buttons: The e2gRuleEngine inference engine deals with values as text, boolean (yes/no or t/f for true/false) and numeric. If a
prompt is provided in the knowledge base for a condition, the prompt type determines the data type. If no prompt is defined, or the value is for an
action that is not also a condition (prompts aren't allowed for actions), clicking one of these buttons determines the data type. If a prompt is defined
later, this setting will be overridden.
4. Goal? check box: If checked, a GOAL statement will be generated for the action in the e2gRuleEngine knowledge base. A condition that is not also an
action may not be defined as a goal, so the check box is only enabled when an action row is selected in the decision table.
Condition/action list value controls:
1. Value list box: This list box contains all of the possible values that may be assigned to the currently selected condition or action. These values are
made available for selection in a combo box when rules are constructed in rule mode.
2. NEW LIST VALUE text field: Typing a value in this field, then pressing the Enter key, adds the value to the end of the value list box. Use Ctrl-V to paste
a text string copied from another document. When the Enter key is pressed, the field is cleared and is ready for the entry of the next value. List value
considerations include:
If the data type is boolean (yes/no or true/false) the only legitimate value list values are true and false. When a true/false prompt type is
specified for a condition, these values are automatically generated in the value list.
In most cases, list values are simply entered as text. When this is done, e2gRuleWriter assumes the value will be used in a rule premise or

12 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

consequent with an equality " = " relational operator, and will enclose the value in double quotation mark delimiters when it is output in a rule.
If the data type is text, and a non-equality relational operator is desired for a condition, the text value must be enclosed in delimiters following
the relational operator. For example: ! "Monday" to indicate that the condition value is not equal to Monday. The allowable relational operators
are: < for strictly less than, <= for less than or equal to, > for strictly greater than, >= for greater than or equal to, ! <> or != for not equal to, :
for one of the values of and !: for none of the values of. The allowable delimiters, which must be used in pairs are: " ", ' ' and [ ]. Three pairs of
delimiters are provided to accommodate embedded delimiters in text values.
If the data type is text, and an equality relational operator is desired for a value that includes embedded delimiters, the " = " relational operator
can be explicitly included and alternate delimiters placed around the list value. For example the list value: the title is "Introduction to Expert
Systems" could be specified as a list value with an equality relational operator as follows: = [the title is "Introduction to Expert Systems"]
If the relational operator for a text value is the " : " (one of the values of) operator or " !: " (none of the values of) operator, multiple comparison
values may be provided in one value list entry, with each value enclosed in delimiters. For example: : "Monday" "Tuesday" "Friday" is
interpreted as allowing the condition to have a value of Monday, Tuesday or Friday, and the rule generated for e2gRuleEngine will be of the
form: condition-name : "Monday" "Tuesday" "Friday" and is interpreted as: one of the values of condition-name is Monday, Tuesday or
Friday.
If the value is for a numeric data type condition or action with an intended equality " = " relational operator in the rules, e2gRuleWriter will output
the value without delimiters when it is output in a rule.
If the value is for a numeric data type condition intended to use a relational operator other than equality when output in a rule, type the relational
operator before the value. For example: < 50 for "strictly less than 50." The allowable relational operators for a numeric comparison are: < for
strictly less than, <= for less than or equal to, > for strictly greater than, >= for greater than or equal to, = for equal to and ! <> or != for not
equal to.
If the value is for a numeric data type condition, and a range of values is to be tested, represent the range as in the following example: >25 ^
<50 which means "greater than 25 and less than 50." The separator character is the caret found on the keyboard as the shifted " 6 " key. When
the e2gRuleEngine knowledge base is generated, this list value will produce two and'd clauses in the rule premise: condition > 25 and
condition < 50. This construct is only available for use with numeric data types.
Relational operators other than " = " make no sense in actions because they cannot be used in e2gRuleEngine rule consequents.
3. Delete selected value from list button: When an entry is selected in the value list box by clicking on the entry, this button is enabled. Clicking the
button deletes the value from the list.
4. Edit selected list value button: When an entry is selected in the value list box by clicking on the entry, this button is enabled. Clicking the button loads
the value into the New list value text field for editing. Once the value has been modified as desired, pressing the Enter key replaces the edited value in
the value list. If the old value has already been assigned to conditions or actions in rules (as discussed in the rule mode details, below), those entries
will be updated with the edited value.
5. Move selected value up button: When an entry is selected in the value list box by clicking on the entry, this button is enabled. New list values are
always added to the end of the value list. To move any value to a new location in the list, select the value and click the button. Each click moves the
item up one position. When it reaches the top of the list, another click moves the item to the end of the list.
Default value controls:
1. DEFAULT VALUE text field: Typing a value, then pressing the Enter key sets the value as the optional default value for the selected condition or action
and e2gRuleWriter will generate a DEFAULT statement for the condition or action in the e2gRuleEngine knowledge base. e2gRuleEngine assigns the
default value to the condition or action if it is unsuccessful in determining a value from a prompt or rule. Use Ctrl-V to paste a text value copied from
another document.
2. Set selected value as DEFAULT button: When an entry is selected in the value list box by clicking on the entry, this button is enabled. Clicking the
button sets the selected value as the optional default value for the selected condition or action.
3. CF Value and Set CF for DEFAULT combo box and button: Selecting a Certainty Factor (CF) value from the combo box and clicking the button
assigns the CF to the current DEFAULT value and will appear after the "@" symbol. If a CF less than 100 is selected before defining a DEFAULT
value, the CF will be immediately assigned. A CF of 100 is implied for the DEFAULT and is never displayed with the default value.
4. Clear current DEFAULT value button: The current default value of this condition or action (shown in the text field with a gray background) is removed.
Condition prompt controls (optional prompts define how data will be requested from the user of an e2gRuleEngine knowledge base, and may be specified
only for conditions):
1. Prompt type radio buttons:
YesNo: Selecting this prompt type inserts true and false in the value list box and defines the data type for the condition or action as boolean
(t/f). An opt-out in the form of an "I don't know/would rather not answer" is included with the Yes/No radio button choices when the prompt is
generated by e2gRuleEngine.
MultChoice: This prompt type will create a multiple choice input in e2gRuleEngine, with each entry in the value list box associated with a radio
button. Includes the "I don't know..." opt-out choice.
ForcedChoice: The forced choice in e2gRuleEngine is identical to the multiple choice input except that the "I don't know/would rather not
answer" option is omitted, so the user must select among the available choices.
Choice: Similar to the MultChoice input with the alternatives presented in a drop down combo box instead of with radio buttons. Includes the "I
don't know.." option.
AllChoice: Similar to the MultChoice option with check boxes instead of radio buttons so that multiple selections can be made. This form of
prompt only makes sense for conditions that have a MAXVALS setting greater than one.
Numeric: This prompt accepts a numeric input with optional sign and decimal point.
2. Numeric RANGE text fields: The acceptable range of a numeric prompt input may be specified by entering numeric values in the text fields labeled
RANGE, Enter. e2gRuleEngine will only accept input values from the prompt that are within the inclusive limits of these values. The RANGE fields are
disabled for prompt types other than numeric.
3. Prompt text field: This field represents the question that will presented to the user of an e2gRuleEngine expert system as part of the prompt.
Alternatively, a text value may be pasted that was copied from another document using Ctrl-V. Press the Enter key after typing or pasting a value.
4. Allow CF Input check box: When this box is checked, a set of radio buttons will be generated as part of the prompt allowing the e2gRuleEngine user
to specify the Certainty Factor (CF) associated with their response.
Decision table load/save controls:
1. Start new decision table button: Restores e2gRuleEngine to its startup configuration, clearing any decision table inputs. An option to save the current
table is provided if it may have been changed since the last save:

2. Load decision table button: Generates a file selection dialog box that allows loading of a file with the decision table (.kbt) extension. A save option
may be displayed before a new table is loaded.
3. Save decision table button: Generates a file selection dialog box that allows saving of the current table as a file with the .kbt extension. The
generated dialog will not allow the table to be stored with any other extension. The decision table is saved as it is configured at the time of the save:
when reloaded, the columns will retain their position and widths and the rules will be renumbered from left to right. Decision tables are always saved as
Unicode (UTF-16) files and must not be edited outside of e2gRuleWriter.
4. Save decision table (CSV) button: Saves the current decision table in Comma Separated Values (CSV) format with the .csv file extension. This format
allows the table to be imported to a spreadsheet so that it may be formatted and printed for documentation purposes. Values from the decision table

13 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

are output one row at a time, with each value enclosed in double quotes and with the values separated by commas. If a double quote character is
embedded in a value, it is replaced with two double quotes. The file will be saved as a Unicode (UTF-16) file if the Unicode option is checked on the
rule mode dialog screen.

Detailed tour of e2gRuleWriter features: rule mode - rule construction

In this section we will examine the process of building rules defined by values associated with the conditions and actions created in the condition/action modes. Rule
mode is entered from condition/action mode by clicking on any rule number at the top of the decision table or on a rule name in the row below the rule numbers.

Rules may be repositioned by dragging and dropping the rule number. A rule column may be resized by dragging the vertical bar between rule numbers left or right.
Rules are output to the knowledge base in left-to-right order of the columns as they appear in the decision table when the knowledge base is generated.

Controls that are not appropriate for use in the current state of the decision table are disabled and "grayed" (presented in a light gray color).

Condition and action values are assigned to rules by double clicking a cell entry in the decision table, then clicking the desired value in the drop down combo box that
opens at the cell. The combo box will contain the values that were entered in the value list for a condition or action while working in condition/action mode.

Rule column controls:

1. Add a rule: A new rule column is created and added at the right end of the data table and is selected as the current rule. The rule number will be one
greater than the largest rule number used so far in working with the current decision table.
2. Insert a rule: A new rule column is created, inserted on the left side of the currently selected rule and becomes the current selection. The rule number
will be one greater than the largest rule number used so far in working with the current decision table.
3. Delete selected rule: The currently selected rule is deleted and the rule on its right becomes the selected rule. If the rightmost rule in the decision
table is deleted, the rightmost remaining rule is selected. One rule always remains in the decision table: if the last remaining rule is deleted, its contents
are cleared and the rule column remains in the table. Rule numbers are not reused during a session working with the same table.
4. Duplicate selected rule: The currently selected rule and all of its contents are replicated at the left of the currently selected rule, then the new rule is
selected. A decision table may require several rules that differ in only a few values. Duplicating the rule, then updating these values, can speed the
data entry process.
5. Tooltips enabled? check box: When this box is selected, a tooltip or hint will be displayed when the mouse pointer hovers near a control for which a
tooltip is available. Checking or clearing the tooltip check box in rule mode also checks or clears the box for condition/action mode.
6. AutoExpand? check box: When this box is selected, the active rule is expanded to the maximum width of the rule title or the longest option value in the
currently selected condition/action combo box. When another rule is selected, or condition/action mode is selected, the column is returned to its default
width. When the AutoExpand check box is deselected, all columns are returned to the e2gRuleWriter default width. When this option is not selected,
rule widths are adjusted by dragging the vertical bar between rule numbers and rules retain their adjusted widths.
Editing rule controls:
1. EDITING text field: Typing a name in this field, then pressing the Enter key, assigns this name to the currently selected rule column. The name will be
displayed below the rule number in the selected column. Ctrl-V may be used to paste a text value copied from another document.
2. CF Value and Set CF for selected action combo box and button: By default, actions are assigned a Certainty Factor of 100%. To have the generated
e2gRuleEngine rule assign a different CF to an action, click the action to select it, choose the desired CF from the combo box then click the Set CF for
selected action button. The CF will appear in the rule action in the format: @ 90 and may be changed at any time. Once again 100% CFs are implied,
and are not displayed in the " @ " format.
Automatic condition stub generation:
1. Condition name list box: When the rule mode is selected, this list box contains the names of all of the condition rows for which at least one condition
value was defined in condition/action mode. If a condition name does not appear in the list, it does not have any defined values. As the first step in
automatic rule generation click each of the condition names to be included in the process while holding down the Ctrl key. All of the selected condition
names will be highlighted.
2. Generate rules for selected conditions button: Automatic rule generation produces rules with condition stubs defined by every combination of the
values of the selected conditions. The new rules generated by this process are displayed in the decision table beginning at the currently selected rule
column. The rule names and action stubs will be blank in the generated rules. If a rule already exists for one of the condition stub combinations it will be

14 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

moved into the rule sequence, not generated. The number of rules produced by automatic rule generation is the product of the number of condition
values for each of the selected conditions. If there are three selected conditions, and each condition has five values, 125 rules will be generated!
Before rule generation begins, the following dialog provides an option to cancel the operation:

If more than 500 rules would be generated, an error dialog is presented, and clicking Cancel terminates the process. If rules are generated that are
not considered necessary in the final decision table, they may be automatically removed with the "incomplete rules" option in the decision table
simplification process since they have no defined actions.
Decision table load/save controls: These buttons and their functions are identical with those defined for the condition/action mode.

Detailed tour of e2gRuleWriter features: rule mode - simplifying and validating rules

e2gRuleWriter provides a comprehensive collection of rule simplification and validation options. It is recommended that the decision table be saved before executing
any of these options. The option selection check box settings are saved with the decision table.

Simplification/validation option check boxes:

1. Remove unused conditions/actions and incomplete rules (no conditions or no actions) check box: Empty condition and action rows are
removed from the table along with rules that have empty condition stubs, empty action stubs or both. Condition and action rows are not removed if a
row name has been entered: these rows may only be deleted in the condition/action mode by using the Delete selected condition/action button.
2. Eliminate irrelevant conditions check box: If a set of rules has identical action stubs and identical condition stubs except for one condition that has
different values across the set of rules, these values are examined to see if all values of the condition are covered by the rules. If so, the condition is
irrelevant and may be eliminated from all of the rules. This creates redundant rules, permitting further decision table simplification.
3. Eliminate redundant rules check box: Rule columns with identical condition and action stubs are redundant, and all but one of the redundant rules will
be removed.
4. Combine decision table rule actions where possible check box: If two rules have identical condition stubs and different action stubs, and the action
stubs involve different actions, the rules may combined into a single rule by integrating the action stubs. If the action stubs involve different values of the
same action, the rules cannot be combined in the decision table, but may be combined when e2gRuleEngine rules are generated.
5. Identify contradictory or possibly contradictory rule actions check box: If two rules have identical condition stubs, specify different action values
for the same action in the action stub and only one value of the action is permitted, they represent a contradiction. However, if the action has a
MAXVALS greater than one (multiple values are allowed), the rules might not be contradictory and this determination cannot be made by
e2gRuleWriter. The program will display an error or warning message depending on which of the two situations is relevant. Rules are not automatically
removed as a result of this check.
Simplification/validation execution:
1. Simplify and error check table button: Clicking this button initiates the requested actions. An option to save the table is provided before table
processing begins, and is highly recommended as data table modifications accomplished by the simplification/validation steps cannot be undone.
2. Summary of simplification and validation actions window: After the Simplify and error check button is clicked, a window opens in which the results
of the requested options are displayed. A title line is displayed for each of the options only if a modification of the decision table has been
accomplished or is recommended by that option. If a permanent record of the simplification actions is desired, the content of this window may be
copied and pasted into a text editor for printing: select the desired text then type Ctrl-C to copy it to the clipboard and Ctrl-V to paste it into a
document.

Detailed tour of e2gRuleWriter features: rule mode - generating an e2gRuleEngine knowledge base

Several options are available for the decision table to e2gRuleEngine knowledge base conversion process. Check box selections are saved with the decision table
and restored when it is reloaded.

Knowledge base options:

1. Combine KB rule conditions where possible check box: Decision table columns may be combined in two ways to form e2gRuleEngine rules when
this option is selected. If there is a set of rules with identical actions and only one condition is used in each of these rules, the rule conditions may be
combined with the "or" logical operator to form a single e2gRuleEngine rule. In addition, selecting this option causes the action stubs in rule columns
with identical condition stubs to be combined into a single e2gRuleEngine rule. The actions will be output with "and" logical operators in the rule
consequent. This option extends the Combine decision table rule actions where possible option of the simplification/validation capability, allowing
rule columns with different values of the same action to be integrated in e2gRuleEngine rules. When rules are combined in the generated knowledge
base, a comment (REM) is inserted in the knowledge base to document the action.
2. Save knowledge base in Unicode format (UTF-16) - default is US-ASCII check box: When this option is selected, the knowledge base file is
produced in Unicode format. This option is required when non-Roman character sets are used in the decision table and is automatically specified if a
translation table is incorporated into the knowledge base. If this option is employed, the Applet tags used to run e2gRuleEngine with the generated
knowledge base must include the following:

<PARAM NAME="ENCODING" VALUE="UTF-16">

If this parameter is not specified, the .kb file will not be recognized as a valid knowledge base, and ERROR 750 will result.
3. Incorporate following translation table into KB check box: If checked, the translation table, used to deliver expert systems in languages other than

15 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

English, identified in the following text field will be added to the knowledge base when it is generated. When this check box is selected the Save
knowledge base in Unicode format (UTF-16)... box is automatically selected.
4. Translation table name, browse text display field and button: Click the browse button to select a translation table to incorporate in the e2gRuleEngine
knowledge base. The use of translation tables is discussed in the e2gRuleEngine documentation. The translation table must have a .txt extension and
the file must be stored in Unicode (UTF-16) format. The complete path to a translation table is stored with the decision table when it is saved, and
restored when it is reloaded. If the decision table (.kbt) file is moved to a different computer the Browse button should be used to reset the location of
the translation table.
Knowledge base generator execution:
1. Display knowledge base button: Clicking this button opens a new window in which the e2gRuleEngine knowledge base corresponding to the decision
table is displayed.
2. Save knowledge base button: Opens a file dialog allowing a knowledge base file name to be selected or specified. The knowledge base file is
required to have a .kb extension. The knowledge base is generated and written to this file.
3. MINCF combo box: The value selected from this drop down list will be written to the generated knowledge base as the MINCF value: the minimum
Certainty Factor an attribute must attain to be considered a fact. The default value is 80%.

e2gRuleWriter examples

Several decision table (.kbt) files are included with the e2gRuleWriter to support experimentation with the software's features. If the software has been
installed as recommended, the decision table and supporting files will be in the same subdirectory as the e2gRuleWriter.jar and e2gRuleEngine.jar files. To
use the examples, load the .kbt file and consider the suggestions provided for each example.

Limited format example (yesno.htm and yesno.kbt):

1. Load yesno.kbt and click any rule number to move to the rule mode then click the Simplify and error check table button.
2. Click the Eliminate irrelevant conditions check box.
3. Compare the simplified table with the original (below) and the summary of simplification actions.
4. Click the Eliminate redundant rules check box then click the Simplify and error check table button.
5. Compare the simplified table with the original (below) and the summary of simplification actions.
6. Reload yesno.kbt and rerun the simplification process with both option check boxes selected.

Auto diagnosis example (auto.htm and auto.kbt):

This is the decision table representation of the "car won't start" example used extensively as an example in these tutorials. Some features to observe:
1. "The gas tank" is used as both a condition and action in this example. This is a typical requirement for intermediate rules such as Rule 5 that
provide, as their action, the condition value for another rule rather than a value for the goal action.
2. Change to the rule mode, and make sure the Combine KB rule conditions... check box is selected then click Display the knowledge base.
Note how Rule 1 and Rule 2 are combined in the generated KB with the use of the "or" logical operator.
3. Uncheck the Combine KB rule conditions... check box and display the knowledge base again to view it in its unsimplified form.
4. Examine the condition mode screen to see the use of a value range for the amount you are willing to spend numeric variable.

16 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Auto diagnosis example - French translation (autofr.htm, autofr.kbt and french16.txt):

1. Load autofr.kbt and click any rule number to move to the rule mode.
2. Click the Incorporate following translation table... check box.
3. Click the Browse button and navigate to french16.txt (the translation table).
4. Examine the rules: using AutoExpand or dragging the column widths with the bar between rule numbers facilitates this activity.
5. Display the e2gRuleEngine knowledge base and examine the way two decision table columns were combined into a single rule with the "or"
logical operator and the incorporation of the french16.txt translation table.
6. It is possible to insert ASCII special characters that are not on the keyboard by typing their decimal codes. Position the cursor at the desired
location in an e2gRuleWriter text field, put the keypad into NumLoc mode then, while holding down the ALT key, type the four digit code for the
ASCII character (the code will begin with a 0). Release the ALT key and the character should appear at the cursor position.
7. Another way to enter special characters in decision table fields is to use a text editor to create the foreign language text, then copy and paste
the text into e2gRuleWriter text fields. Copy text by selecting it then pressing Ctrl-C. Paste it at the current cursor position by pressing Ctrl-V.
This process was used to build this example and the Chinese example that follows. Because specialized editors that facilitate text entry in
specific non-English languages are often available, this technique is recommended.

17 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

Weather example - Chinese translation (weatherch.htm, weatherch.kbt and chinese16.txt):

This example incorporates a partially translated table (chinese16.txt) and its examination in a similar fashion to the French translation example above is
recommended. Use of an external text editor to construct the text for a non-Roman language application like this one is a necessity. The text for this
example was machine translated by Yahoo's Babel Fish, and no claims are made for the accuracy of this translation.

Wine choice example (wine.htm and wine.kbt):

The final example incorporates more conditions, actions and rules than earlier examples.
1. You will need to expand the display or scroll the decision table vertically to view the actions because of the large number of conditions.
2. This example makes extensive use of the "one of the values of" option for condition values as illustrated in the screen diagram below. The
autoExpand? option is especially useful in this example because of the wide decision table columns required to show the full text of the
condition/action options. Note also that the combo box can scroll if the option list is lengthy.

18 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

3. The wine example provides a good opportunity to illustrate the dangers of the automatic rule generation process. Select the rule mode by
clicking on any rule number, then select all of the conditions in the Automatic rule generation list box (click the 1st condition then, while holding
down the shift key, click the last condition). When you click the Generate rules for selected conditions button, you will see an error message
informing you that over 150,000 rules would be generated!

4. Click the Cancel key to cancel the operation, then experiment with some other condition selections for rule generation. e2gRuleEngine will allow
up to 500 rules to be generated, but even this number is probably more than you will want to consider. If the number of rules to be generated is
less than or equal to 500, the count is displayed in a dialog box with an option to continue or cancel the operation.

e2gRuleWriter operating system and user interface considerations

The example screens in this document represent the appearance of e2gRuleWriter when run under the Microsoft Windows operating system. The minimum
screen resolution running under Windows is 800 x 600 pixels, and the program will open at this resolution. The application may be expanded to show more
decision table rows and columns using the "maximize" button on the frame or by dragging the frame border.

If e2gRuleWriter is run using any operating system other than Windows, the user interface is set to the Java Cross Platform User Interface ("Metal") and the
minimum screen resolution will be set to 1024 x 768 pixels to accommodate larger default text sizes and components. When run with this user interface, the
screen will differ slightly in appearance from this document's earlier examples as illustrated below:

19 of 20 8.6.2016 9:50
e2gRuleWriter decision table software: e2gRuleEngine knowledge base... http://www.expertise2go.com/e2g3g/e2g3gdoc/e2gRuleWriterRef.htm

If you are running e2gRuleWriter on a Windows platform, and would like to work in the cross platform interface, start the program with the following command
line:

java -jar e2gRuleWriter.jar -M

If you have a problem with the text displayed when you start the program this way (may be caused by missing fonts) try starting the program with the
following command that forces the use of Arial Unicode MS as Java's Dialog font:

java -jar e2gRuleWriter.jar -MM

[Index][Module 1][Module 2][Module 3][Module 4][Module 5][Module 6][Module 7][Module 7a][Module 8][Module 9][Reference]

Copyright © 2009-2011 by eXpertise2Go.com. All rights reserved.

20 of 20 8.6.2016 9:50