e2gLite Free Expert System Shell

[Home: Demos and Tutorials][e2gLite Mini-Course Index]

Building and Using Expert Systems: a Mini-Course Introducing the e2gLite Expert System Shell

Reference: Tables summarizing e2gLite's commands and error messages.

Reference Index

Example HTML to incorporate the e2gLite applet in a Web page

e2gLite applet parameters

e2gLite button codes passed to Javascript

e2gLite public methods accessible from Javascript

Example knowledge base

Knowledge base format

Relational operators

Text Customizeable with TRANSLATE Commands

Error codes

Knowledge base trace output

Knowledge base analysis output

Knowledge base dump output

HTML to incorporate the e2gLite Applet in a Web page

<APPLET CODE="e2glite.e2g.class" ARCHIVE="e2glite.jar" WIDTH=450 HEIGHT=300 MAYSCRIPT> <PARAM NAME="KBURL" VALUE="auto.kb"> (optional <PARAM> tags) . . . Java capable browser required! <BR>Free Java capable browsers are available from: <A HREF="http://www.netscape.com">Netscape</A> and <A HREF="http://www.opera.com/download">Opera</A> </APPLET>

Notes

The APPLET CODE and ARCHIVE parameters are case sensitive and are lower case for e2gLite.
If the e2glite.jar file is not located in the same subdirectory as the HTML document that declares the applet, the ARCHIVE parameter must provide a path to the file.
The MAYSCRIPT parameter is only included when e2gLite will call a Javascript function (Version 3.x). See the JSFUNCTION parameter below.

HTML to incorporate the e2gLite Applet in a Web page
<APPLET CODE="e2glite.e2g.class" ARCHIVE="e2glite.jar" WIDTH=450 HEIGHT=300 MAYSCRIPT> <PARAM NAME="KBURL" VALUE="auto.kb"> (optional <PARAM> tags) . . . Java capable browser required! <BR>Free Java capable browsers are available from: <A HREF="http://www.netscape.com">Netscape</A> and <A HREF="http://www.opera.com/download">Opera</A> </APPLET>
Notes	The APPLET CODE and ARCHIVE parameters are case sensitive and are lower case for e2gLite. If the e2glite.jar file is not located in the same subdirectory as the HTML document that declares the applet, the ARCHIVE parameter must provide a path to the file. The MAYSCRIPT parameter is only included when e2gLite will call a Javascript function (Version 3.x). See the JSFUNCTION parameter below.

e2gLite Applet Parameters

Name Values Default Examples

KBURL Knowledge base name, path from directory containing document that loaded applet or complete URL locating the KB No default: required parameter <PARAM NAME="KBURL" VALUE="auto.kb">
(KB is in same directory as HTML document)
<PARAM NAME="KBURL" VALUE="/kb/auto.kb">
(file is in kb subdirectory below directory containing HTML document)
<PARAM NAME="KBURL" VALUE="http://www.mysite.com/kb/auto.kb">

APPTITLE Any character string: 1st line of title on start screen Blank <PARAM NAME="APPTITLE" VALUE="Auto Diagnosis">

APPSUBTITLE Any character string: 2nd line of title on start screen Blank <PARAM NAME="APPSUBTITLE" VALUE="by Ace Auto Repair">

BGCOLOR Applet background color specified as hex RGB color as used in HTML #FFEE33 (pale yellow) <PARAM NAME="BGCOLOR" VALUE="#00FF00"> (green)
<PARAM NAME="BGCOLOR" VALUE="#0000FF"> (blue)

TITLECOLOR APPTITLE and APPSUBTITLE text color specified as hex RGB color as used in HTML #000000 (black) <PARAM NAME="TITLECOLOR" VALUE="#FFFFFF"> (white)

PROMPTCOLOR PROMPT text color specified as hex RGB color as used in HTML #000000 (black) <PARAM NAME="PROMPTCOLOR" VALUE="#FFFFFF"> (white)

DEBUG If TRUE, debug window is created when KB is started FALSE <PARAM NAME="DEBUG" VALUE="TRUE">

STARTBUTTON Text on button that starts consultation "Start the consultation" <PARAM NAME="STARTBUTTON" VALUE="Help me!">

LOADMSG Message displayed on opening screen while knowledge base is loaded "Loading knowledge base..." <PARAM NAME="LOADMSG" VALUE="Chargeant la base de connaissance...">

PROMPTFIRST If TRUE use PROMPT for an attribute before backward chaining TRUE <PARAM NAME="PROMPTFIRST" VALUE="false">

JSFUNCTION (Version 3.x) Provides the name of a Javascript function to be called each time an e2gLite button is clicked null (no function is called) <PARAM NAME="JSFUNCTION" VALUE="buttonPush">

Notes

PARAM NAMEs are case sensitive and are always defined in upper case in e2gLite.
e2gLite PARAM VALUEs are not case sensitive unless they represent file names for an operating system that treats file names as case sensitive (Unix variants).
KBURL is the only required parameter.
If the KBURL value does not start with http://, the value will be appended to the directory containing the HTML file that loaded the e2glite applet.
The knowledge base must be loaded from the same Web site as the applet archive (e2glite.jar). This is a security requirement of Java applets.
When the JSFUNCTION parameter is used, MAYSCRIPT must be included in the <applet> tag.

e2gLite Applet Parameters
Name	Values	Default	Examples
KBURL	Knowledge base name, path from directory containing document that loaded applet or complete URL locating the KB	No default: required parameter	<PARAM NAME="KBURL" VALUE="auto.kb"> (KB is in same directory as HTML document) <PARAM NAME="KBURL" VALUE="/kb/auto.kb"> (file is in kb subdirectory below directory containing HTML document) <PARAM NAME="KBURL" VALUE="http://www.mysite.com/kb/auto.kb">
APPTITLE	Any character string: 1st line of title on start screen	Blank	<PARAM NAME="APPTITLE" VALUE="Auto Diagnosis">
APPSUBTITLE	Any character string: 2nd line of title on start screen	Blank	<PARAM NAME="APPSUBTITLE" VALUE="by Ace Auto Repair">
BGCOLOR	Applet background color specified as hex RGB color as used in HTML	#FFEE33 (pale yellow)	<PARAM NAME="BGCOLOR" VALUE="#00FF00"> (green) <PARAM NAME="BGCOLOR" VALUE="#0000FF"> (blue)
TITLECOLOR	APPTITLE and APPSUBTITLE text color specified as hex RGB color as used in HTML	#000000 (black)	<PARAM NAME="TITLECOLOR" VALUE="#FFFFFF"> (white)
PROMPTCOLOR	PROMPT text color specified as hex RGB color as used in HTML	#000000 (black)	<PARAM NAME="PROMPTCOLOR" VALUE="#FFFFFF"> (white)
DEBUG	If TRUE, debug window is created when KB is started	FALSE	<PARAM NAME="DEBUG" VALUE="TRUE">
STARTBUTTON	Text on button that starts consultation	"Start the consultation"	<PARAM NAME="STARTBUTTON" VALUE="Help me!">
LOADMSG	Message displayed on opening screen while knowledge base is loaded	"Loading knowledge base..."	<PARAM NAME="LOADMSG" VALUE="Chargeant la base de connaissance...">
PROMPTFIRST	If TRUE use PROMPT for an attribute before backward chaining	TRUE	<PARAM NAME="PROMPTFIRST" VALUE="false">
JSFUNCTION	(Version 3.x) Provides the name of a Javascript function to be called each time an e2gLite button is clicked	null (no function is called)	<PARAM NAME="JSFUNCTION" VALUE="buttonPush">
Notes	PARAM NAMEs are case sensitive and are always defined in upper case in e2gLite. e2gLite PARAM VALUEs are not case sensitive unless they represent file names for an operating system that treats file names as case sensitive (Unix variants). KBURL is the only required parameter. If the KBURL value does not start with http://, the value will be appended to the directory containing the HTML file that loaded the e2glite applet. The knowledge base must be loaded from the same Web site as the applet archive (e2glite.jar). This is a security requirement of Java applets. When the JSFUNCTION parameter is used, MAYSCRIPT must be included in the <applet> tag.

Button Codes Provided by e2gLite

Button Clicked Value

Start the consultation * 1

Submit your response 2

Why Ask? 3

Return (from Why Ask?) 4

Explain 5

Restart 6

* May be renamed by the STARTBUTTON applet parameter.
Other buttons may be renamed by knowledge base TRANSLATE commands.

Button Codes Provided by e2gLite
Button Clicked	Value
Start the consultation *	1
Submit your response	2
Why Ask?	3
Return (from Why Ask?)	4
Explain	5
Restart	6
* May be renamed by the STARTBUTTON applet parameter. Other buttons may be renamed by knowledge base TRANSLATE commands.

e2gLite Methods Accessible from Javascript

Method Name Value Returned and Value Type Arguments and Argument Types

getAttrCF(attrName,valIx) (Float) Confidence factor associated with specific instance of named attribute (0.0 if doesn't exist or net yet determined) (String) attrName: name of attribute
(Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes

getAttrCount( ) (Integer) Number of attributes defined in the knowledge base

getAttrIx(attrName,valIx) (Integer) Index of attribute associated with specific instance of named attribute (0 if doesn't exist) (String) attrName: name of attribute
(Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes

getAttrName(attrIx) (String) Name of attribute (zero length string if doesn't exist) (Integer) attrIx: index of attribute

getAttrValue(attrIx) (String) String representation of current value (0 length string if doesn't exist or not yet determined) (Integer) attrIx: index of attribute

getAttrValue(attrName,valIx) (String) String representation of current value (0 length string if doesn't exist or not yet determined) (String) attrName: name of attribute
(Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes

getMaxValues(attrName) (Integer) Number of values (instances) allowed for named attribute (0 if doesn't exist) (String) attrName: name of attribute

getGoalAttr( ) (Integer) Index of current goal or subgoal attribute (0 if inferencing is finished)

Note 1: e2gLite attributes are identified with a name and index number that begins at 1. Every attribute value has a unique index, but all values of multivalued attributes are stored with the same name.
Note 2: A value for an attribute will not be returned until the value is known with a confidence factor at least equal to the minimum CF.

e2gLite Methods Accessible from Javascript
Method Name	Value Returned and Value Type	Arguments and Argument Types
getAttrCF(attrName,valIx)	(Float) Confidence factor associated with specific instance of named attribute (0.0 if doesn't exist or net yet determined)	(String) attrName: name of attribute (Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes
getAttrCount( )	(Integer) Number of attributes defined in the knowledge base
getAttrIx(attrName,valIx)	(Integer) Index of attribute associated with specific instance of named attribute (0 if doesn't exist)	(String) attrName: name of attribute (Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes
getAttrName(attrIx)	(String) Name of attribute (zero length string if doesn't exist)	(Integer) attrIx: index of attribute
getAttrValue(attrIx)	(String) String representation of current value (0 length string if doesn't exist or not yet determined)	(Integer) attrIx: index of attribute
getAttrValue(attrName,valIx)	(String) String representation of current value (0 length string if doesn't exist or not yet determined)	(String) attrName: name of attribute (Integer) valIx: instance (between 1 and max values) of named attribute to allow for multi-valued attributes
getMaxValues(attrName)	(Integer) Number of values (instances) allowed for named attribute (0 if doesn't exist)	(String) attrName: name of attribute
getGoalAttr( )	(Integer) Index of current goal or subgoal attribute (0 if inferencing is finished)
Note 1: e2gLite attributes are identified with a name and index number that begins at 1. Every attribute value has a unique index, but all values of multivalued attributes are stored with the same name. Note 2: A value for an attribute will not be returned until the value is known with a confidence factor at least equal to the minimum CF.

Example e2gLite Knowledge Base

REM Test knowledge base RULE [Is the battery dead?] If [the result of switching on the headlights] = "nothing happens" or [the result of trying the starter] = "nothing happens" Then [the recommended action] = "recharge or replace the battery" RULE [Is the car out of gas?] If [the gas tank] = "empty" Then [the recommended action] = "refuel the car" RULE [Is the battery weak?] If [the result of trying the starter] : "the car cranks slowly" "the car cranks normally" and [the headlights dim when trying the starter] = true and [the amount you are willing to spend on repairs] > 24.99 Then [the recommended action] = "recharge or replace the battery" RULE [Is the car flooded?] If [the result of trying the starter] = "the car cranks normally" and [a gas smell] = "present when trying the starter" Then [the recommended action] = "wait 10 minutes, then restart flooded car" RULE [Is the gas tank empty?] If [the result of trying the starter] = "the car cranks normally" and [a gas smell] = "not present when trying the starter" Then [the gas tank] = "empty" @ 90 PROMPT [the result of trying the starter] Choice CF "What happens when you turn the key to try to start the car?" "the car cranks normally" "the car cranks slowly" "nothing happens" PROMPT [a gas smell] MultChoice CF "The smell of gasoline is:" "present when trying the starter" "not present when trying the starter" PROMPT [the result of switching on the headlights] MultChoice CF "The result of switching on the headlights is:" "they light up" "nothing happens" PROMPT [the headlights dim when trying the starter] YesNo CF "Do the headlights dim when you try the starter with the lights on?" PROMPT [the gas tank] MultChoice CF "According to the fuel gauge, the gas tank is:" "empty" "not empty" PROMPT [the amount you are willing to spend on repairs] Numeric CF "How much are you willing to spend on repairs? (enter value 0->500)" "0" "500.0" GOAL [the recommended action] MINCF 80

Example e2gLite Knowledge Base
REM Test knowledge base RULE [Is the battery dead?] If [the result of switching on the headlights] = "nothing happens" or [the result of trying the starter] = "nothing happens" Then [the recommended action] = "recharge or replace the battery" RULE [Is the car out of gas?] If [the gas tank] = "empty" Then [the recommended action] = "refuel the car" RULE [Is the battery weak?] If [the result of trying the starter] : "the car cranks slowly" "the car cranks normally" and [the headlights dim when trying the starter] = true and [the amount you are willing to spend on repairs] > 24.99 Then [the recommended action] = "recharge or replace the battery" RULE [Is the car flooded?] If [the result of trying the starter] = "the car cranks normally" and [a gas smell] = "present when trying the starter" Then [the recommended action] = "wait 10 minutes, then restart flooded car" RULE [Is the gas tank empty?] If [the result of trying the starter] = "the car cranks normally" and [a gas smell] = "not present when trying the starter" Then [the gas tank] = "empty" @ 90 PROMPT [the result of trying the starter] Choice CF "What happens when you turn the key to try to start the car?" "the car cranks normally" "the car cranks slowly" "nothing happens" PROMPT [a gas smell] MultChoice CF "The smell of gasoline is:" "present when trying the starter" "not present when trying the starter" PROMPT [the result of switching on the headlights] MultChoice CF "The result of switching on the headlights is:" "they light up" "nothing happens" PROMPT [the headlights dim when trying the starter] YesNo CF "Do the headlights dim when you try the starter with the lights on?" PROMPT [the gas tank] MultChoice CF "According to the fuel gauge, the gas tank is:" "empty" "not empty" PROMPT [the amount you are willing to spend on repairs] Numeric CF "How much are you willing to spend on repairs? (enter value 0->500)" "0" "500.0" GOAL [the recommended action] MINCF 80

e2gLite Knowledge Base Format

Command Description Examples

REM Comment line REM Diagnose why an automobile won't start

RULE Begins a rule: followed by rule description in square brackets RULE [Is the battery dead?]
If [the result of switching on the headlights] = "nothing happens" or
[the result of trying the starter] = "nothing happens"
Then [the problem] = "the battery is discharged or dead" @ 90 and
[the recommended action] = "recharge or replace the battery"

   If Begins rule premise: followed by first premise clause (attribute name in square brackets + relational operator + comparison quantity), optional logical operator and/or (if there is another premise clause)

   Then Begins rule consequent: followed by first consequent clause (attribute name in square brackets + = + value + optional confidence/certainty factor: @ xxx where xxx represents a percentage and defaults to 100), optional and (required if there is another consequent clause)

PROMPT Defines format of user request for an attribute's value if the inference engine is allowed to request it.

Prompt types:
  YesNo (radio button true/false (Boolean) input)
  MultChoice (text string radio button input)
  Choice (text string drop-down list input)
  AllChoice (text string check box multiple inputs: See MAXVALS)
  Numeric (range-checked numeric input)

Optional CF parameter generates request for confidence factor input

Question to ask user is on the next line in single or double quotes.

Response choices (if required by the PROMPT) are on successive lines in quotes

For the Numeric PROMPT, the minimum and maximum acceptable values are on successive lines in quotes PROMPT [the stoplights work] YesNo CF
"Do the stoplights light when you step on the brake?"

PROMPT [the gas tank] MultChoice CF
"According to the fuel gauge, the gas tank is:"
"empty"
"not empty"

PROMPT [the result of trying the starter] Choice CF
"What happens when you turn the key to try to start the car?"
"the car cranks normally"
"the car cranks slowly"
"nothing happens"

PROMPT [an acceptable time to pick up your car] Choice
What time may we pick up your car? (Check up to three)
"7 - 9 AM"
"9 AM -12 PM"
"12 - 3 PM"
"3 - 7 PM"

PROMPT [the amount you are willing to spend on repairs] Numeric CF
"How much are you willing to spend on repairs? (enter value 0->500)"
"0"
"500.0"

GOAL Finding a value (or values) for the goal variables is the objective of an expert system consultation. When all GOALs are resolved (or it is determined they cannot be resolved) the consultation ends. At least one GOAL is required. GOAL [the problem]

MINCF (Optional) The minimum confidence/certainty factor to accept a value as a fact. Default is 80.0 MINCF 70

MAXVALS (Optional) The maximum number of different values one attribute can have. Default is one. Attributes with AllChoice PROMPTs should have a MAXVALS greater than one. MAXVALS [an acceptable time to pick up your car] 3

DEFAULT (Optional) Sets value of an attribute if the inference engine fails to determine its value from PROMPTs or RULEs. DEFAULT [the gas tank] = "not empty"
DEFAULT [the stoplights work] = true
DEFAULT [the amount you are willing to spend on repairs] = 50

TRANSLATE (Optional) Customizes e2gLite text output. Also supports knowledge base translation to languages other than English. See Text Customizeable with Translate Commands for a list of text items that may be customized. TRANSLATE B_SUBMIT = "Soumettez votre réponse"
TRANSLATE TR_YES = "Oui"
TRANSLATE TR_RULE = "RÈGLE:"

Notes

All RULEs must be defined before any other command (except REM) is used.
Attribute names must be enclosed in square brackets.
Text strings representing attribute values must be enclosed in matching single or double quotes.
Commands, attribute names, and text values are not case-sensitive.
Spaces within square brackets or quotes are significant. Spaces between other elements are not significant.
Multiple rule premise clauses may be connected with either And or Or but not both in the same rule.

e2gLite Knowledge Base Format
Command	Description	Examples
REM	Comment line	REM Diagnose why an automobile won't start
RULE	Begins a rule: followed by rule description in square brackets	RULE [Is the battery dead?] If [the result of switching on the headlights] = "nothing happens" or [the result of trying the starter] = "nothing happens" Then [the problem] = "the battery is discharged or dead" @ 90 and [the recommended action] = "recharge or replace the battery"
If	Begins rule premise: followed by first premise clause (attribute name in square brackets + relational operator + comparison quantity), optional logical operator and/or (if there is another premise clause)
Then	Begins rule consequent: followed by first consequent clause (attribute name in square brackets + = + value + optional confidence/certainty factor: @ xxx where xxx represents a percentage and defaults to 100), optional and (required if there is another consequent clause)
PROMPT	Defines format of user request for an attribute's value if the inference engine is allowed to request it. Prompt types: YesNo (radio button true/false (Boolean) input) MultChoice (text string radio button input) Choice (text string drop-down list input) AllChoice (text string check box multiple inputs: See MAXVALS) Numeric (range-checked numeric input) Optional CF parameter generates request for confidence factor input Question to ask user is on the next line in single or double quotes. Response choices (if required by the PROMPT) are on successive lines in quotes For the Numeric PROMPT, the minimum and maximum acceptable values are on successive lines in quotes	PROMPT [the stoplights work] YesNo CF "Do the stoplights light when you step on the brake?" PROMPT [the gas tank] MultChoice CF "According to the fuel gauge, the gas tank is:" "empty" "not empty" PROMPT [the result of trying the starter] Choice CF "What happens when you turn the key to try to start the car?" "the car cranks normally" "the car cranks slowly" "nothing happens" PROMPT [an acceptable time to pick up your car] Choice What time may we pick up your car? (Check up to three) "7 - 9 AM" "9 AM -12 PM" "12 - 3 PM" "3 - 7 PM" PROMPT [the amount you are willing to spend on repairs] Numeric CF "How much are you willing to spend on repairs? (enter value 0->500)" "0" "500.0"
GOAL	Finding a value (or values) for the goal variables is the objective of an expert system consultation. When all GOALs are resolved (or it is determined they cannot be resolved) the consultation ends. At least one GOAL is required.	GOAL [the problem]
MINCF	(Optional) The minimum confidence/certainty factor to accept a value as a fact. Default is 80.0	MINCF 70
MAXVALS	(Optional) The maximum number of different values one attribute can have. Default is one. Attributes with AllChoice PROMPTs should have a MAXVALS greater than one.	MAXVALS [an acceptable time to pick up your car] 3
DEFAULT	(Optional) Sets value of an attribute if the inference engine fails to determine its value from PROMPTs or RULEs.	DEFAULT [the gas tank] = "not empty" DEFAULT [the stoplights work] = true DEFAULT [the amount you are willing to spend on repairs] = 50
TRANSLATE	(Optional) Customizes e2gLite text output. Also supports knowledge base translation to languages other than English. See Text Customizeable with Translate Commands for a list of text items that may be customized.	TRANSLATE B_SUBMIT = "Soumettez votre réponse" TRANSLATE TR_YES = "Oui" TRANSLATE TR_RULE = "RÈGLE:"
Notes	All RULEs must be defined before any other command (except REM) is used. Attribute names must be enclosed in square brackets. Text strings representing attribute values must be enclosed in matching single or double quotes. Commands, attribute names, and text values are not case-sensitive. Spaces within square brackets or quotes are significant. Spaces between other elements are not significant. Multiple rule premise clauses may be connected with either And or Or but not both in the same rule.

e2gLite Relational Operators

Operator Interpretation Formatted Rule Translation Example unformatted/formatted premise clause output

= equal to is
(NOTE: With true Boolean attributes, the relational operator and comparison quantity is omitted from translation) [the result of trying the starter] = "nothing happens"
the result of trying the starter is nothing happens
[the headlights dim when the starter cranks] = true
the headlights dim when the starter cranks
[the headlights dim when the starter cranks] = false
the headlights dim when the starter cranks is false

< strictly less than is less than [the age of the battery in months] < 24
the age of the battery in months is less than 24

> strictly greater than is greater than [the age of the battery in months] > 12
the age of the battery in months is greater than 12

! not equal to is not [the gas tank] ! "empty"
the gas tank is not empty

: equals any of
(NOTE: use only with text string data) is value, value,... or value [the battery brand] : "Delco" "Mopar" "Interstate"
the battery brand is Delco, Mopar or Interstate

Notes

One or more optional spaces may be included on either side of the relational operator.
The translated value of the relational operator should be considered when creating attribute names and values to improve rule readability.
With the exception of : (equals any of), the value of a multivalued attribute with the highest CF is used to evaluate an expression. With the equals any of operator, the expression is considered true if any value of the attribute with a certainty factor equal to or greater than MINCF matches any value in the list of comparison quantities.

e2gLite Relational Operators
Operator	Interpretation	Formatted Rule Translation	Example unformatted/formatted premise clause output
=	equal to	is (NOTE: With true Boolean attributes, the relational operator and comparison quantity is omitted from translation)	[the result of trying the starter] = "nothing happens" the result of trying the starter is nothing happens [the headlights dim when the starter cranks] = true the headlights dim when the starter cranks [the headlights dim when the starter cranks] = false the headlights dim when the starter cranks is false
<	strictly less than	is less than	[the age of the battery in months] < 24 the age of the battery in months is less than 24
>	strictly greater than	is greater than	[the age of the battery in months] > 12 the age of the battery in months is greater than 12
!	not equal to	is not	[the gas tank] ! "empty" the gas tank is not empty
:	equals any of (NOTE: use only with text string data)	is value, value,... or value	[the battery brand] : "Delco" "Mopar" "Interstate" the battery brand is Delco, Mopar or Interstate
Notes	One or more optional spaces may be included on either side of the relational operator. The translated value of the relational operator should be considered when creating attribute names and values to improve rule readability. With the exception of : (equals any of), the value of a multivalued attribute with the highest CF is used to evaluate an expression. With the equals any of operator, the expression is considered true if any value of the attribute with a certainty factor equal to or greater than MINCF matches any value in the list of comparison quantities.

Text Customizeable with TRANSLATE Commands

REM Button text TRANSLATE B_SUBMIT = "Submit your response" TRANSLATE B_EXPLAIN = "Explain" TRANSLATE B_WHYASK = "Why ask?" TRANSLATE B_RESTART = "Restart" TRANSLATE B_RETURN = "Return" REM Message text TRANSLATE TR_KB = "Knowledge base:" TRANSLATE TR_NORESP = "I don't know/would rather not answer" TRANSLATE TR_HOWCONF = "How confident do you feel about your response?" TRANSLATE TR_LOWCONF = "Very uncertain (50%)" TRANSLATE TR_HICONF = "Very certain (100%)" TRANSLATE TR_YES = "yes" TRANSLATE TR_NO = "no" TRANSLATE TR_FALSE = "false" TRANSLATE TR_RESULTS = "FINAL RESULTS:" TRANSLATE TR_MINCF = "Minimum confidence factor for accepting a value as a fact:" TRANSLATE TR_NOTDETERMINED = "could not be determined" TRANSLATE TR_ISRESULT = "is:" TRANSLATE TR_WITH = "with" TRANSLATE TR_CONF = "% confidence" TRANSLATE TR_ALLGOALS = "all conclusions" TRANSLATE TR_VALUE = "Value" TRANSLATE TR_OF = "of" TRANSLATE TR_THISRULE = "Rule below fired at CF=" TRANSLATE TR_RULEASGN = "and assigned the value" TRANSLATE TR_TOFIND = "To find" TRANSLATE TR_AVALUE = "a value for" TRANSLATE TR_ISNEEDED = "is needed to try this rule:" TRANSLATE TR_RULE = "RULE:" TRANSLATE TR_IF = "IF:" TRANSLATE TR_THEN = "THEN:" TRANSLATE TR_AND = "and" TRANSLATE TR_OR = "or" TRANSLATE TR_EQUAL = "is" TRANSLATE TR_LESSTHAN = "is less than" TRANSLATE TR_GREATER = "is greater than" TRANSLATE TR_NOTEQUAL = "is not" TRANSLATE TR_VALUEFOR = "A value for:" TRANSLATE TR_FOUND = "has been determined" TRANSLATE TR_NOTFOUND = "has not yet been determined" TRANSLATE TR_WASINPUT = "was input with " TRANSLATE TR_DETERMINED = "Determined" TRANSLATE TR_IS = "is" TRANSLATE TR_FROM = "from:" TRANSLATE TR_DEFAULTED = "was set by default with"

Notes

The default value for each customizeable item is shown in the format of a TRANSLATE command for illustration only. Only those text strings that are to be changed from their default value should be redefined with TRANSLATE commands. To deliver a non-English version of a knowledge base, all items should be translated. An example in French is included in Module 7.
See also the STARTBUTTON and LOADMSG Applet Parameters.

e2gLite Error Codes

Error Meaning Comment

150 Initialization error Internal error while trying to load KB

210 Too many GOAL statements Program capacity exceeded

220 No GOAL statements At least one goal is required

230 Too many attributes Program capacity exceeded

240 Too many unknown attributes Program capacity exceeded

250 Too many RULEs Program capacity exceeded

260 Too many RULE permise clauses Program capacity exceeded

270 Too many RULE consequent clauses Program capacity exceeded

280 Space for RULE data exceeded Program capacity exceeded

290 Too many PROMPTS Program capacity exceeded

300 Space for PROMPT data exceeded Program capacity exceeded

400 Expected a delimited attribute name or value at this location

410 Unexpected end of knowledge base Encountered end of KB file while processing a KB component

420 Expected an IF statement at this location Next non-blank line after RULE must begin with IF

430 Expected an attribute name at this location

440 Expected a value at this location

450 Expected a THEN statement at this location The premise clause on the previous line did not end with And or Or

460 Expected a relational operator (= < > ! :) at this location

470 Expected a logical operator (and/or) at this location Using the "=" instead of ":" relational operator with multiple values is one source of this error

480 Expected an assignment operator (=) at this location

490 Expected AND at this location

500 Expected numeric confidence factor between 1 and 100 at this location

510 Expected numeric MAXVALS value at this location

520 Expected PROMPT type (YesNo, MultChoice,...) at this location

530 A translation value with this name does not exist

540 An attribute with this name does not exist Usually caused by defining a PROMPT with an attribute name never used in a RULE

600 Attempt to change attribute type Use in an earlier RULE defined this attribute as a different data type

610 Not a legitimate numeric value Illegal characters or out of range

620 Bad numeric value in MAXVALS statement

630 Logical operator not recognized

700 PROMPT type not recognized

750 Knowledge base command not recognized

760 Unrecognized data type

800 Missing or empty knowledge base

Notes

Most numbered errors are detected when the knowledge base is loaded. Loading aborts when the first error is detected and the line containing the error is displayed in the DEBUG window if running in DEBUG mode. (See e2gLite Applet Parameters.)

e2gLite Error Codes
Error	Meaning	Comment
150	Initialization error	Internal error while trying to load KB
210	Too many GOAL statements	Program capacity exceeded
220	No GOAL statements	At least one goal is required
230	Too many attributes	Program capacity exceeded
240	Too many unknown attributes	Program capacity exceeded
250	Too many RULEs	Program capacity exceeded
260	Too many RULE permise clauses	Program capacity exceeded
270	Too many RULE consequent clauses	Program capacity exceeded
280	Space for RULE data exceeded	Program capacity exceeded
290	Too many PROMPTS	Program capacity exceeded
300	Space for PROMPT data exceeded	Program capacity exceeded
400	Expected a delimited attribute name or value at this location
410	Unexpected end of knowledge base	Encountered end of KB file while processing a KB component
420	Expected an IF statement at this location	Next non-blank line after RULE must begin with IF
430	Expected an attribute name at this location
440	Expected a value at this location
450	Expected a THEN statement at this location	The premise clause on the previous line did not end with And or Or
460	Expected a relational operator (= < > ! :) at this location
470	Expected a logical operator (and/or) at this location	Using the "=" instead of ":" relational operator with multiple values is one source of this error
480	Expected an assignment operator (=) at this location
490	Expected AND at this location
500	Expected numeric confidence factor between 1 and 100 at this location
510	Expected numeric MAXVALS value at this location
520	Expected PROMPT type (YesNo, MultChoice,...) at this location
530	A translation value with this name does not exist
540	An attribute with this name does not exist	Usually caused by defining a PROMPT with an attribute name never used in a RULE
600	Attempt to change attribute type	Use in an earlier RULE defined this attribute as a different data type
610	Not a legitimate numeric value	Illegal characters or out of range
620	Bad numeric value in MAXVALS statement
630	Logical operator not recognized
700	PROMPT type not recognized
750	Knowledge base command not recognized
760	Unrecognized data type
800	Missing or empty knowledge base
Notes	Most numbered errors are detected when the knowledge base is loaded. Loading aborts when the first error is detected and the line containing the error is displayed in the DEBUG window if running in DEBUG mode. (See e2gLite Applet Parameters.)

Knowledge Base Trace Output

Statement Description Examples (from various demo KBs)

STARTING NEW CONSULTATION WITH THE FOLLOWING GOALS: List of all attributes named by GOAL statements >the recommendation

Minimum confidence factor for accepting a value as a fact: Provided after the list of goals when a new consultation begins. 80%

TRYING RULE: The rule description for the rule the inference engine is about to consider. Is it going to rain?

>Add to goal stack Provides the name of an attribute added to the goal stack. A value of this attribute is needed to evaluate the rule the inference engine last considered leading to an Unknown status. precipitation

>Rule status after evaluation is: The status of the rule the inference engine just considered: possible values are Unknown, True/fired, False and Failed. Unknown

PROMPT FOR: A PROMPT will be tried to obtain a value for the named attribute. precipitation

*Prompt assigned: The value returned from the last issued PROMPT followed by the confidence with which the value was input. precipitation=expected (100.0%)

*Rule assigned: A rule has fired and the consequent assigned the specified value to the named attribute with the confidence level shown in parentheses. the recommendation=wear a raincoat (100.0%)

*Default assigned: The value specified in a knowledge base DEFAULT command was assigned to the named attribute with the confidence level shonw in parentheses after the inference engine failed to resolve its value. the recommendation=wear whatever you want to (90.0%)

FORWARD CHAIN: The value of an attribute has been determined from a PROMPT, because a rule fired or from a DEFAULT with sufficient confidence to be considered a fact. It will now be used to evaluate relevant rules in the knowledge base. precipitation=expected (100.0%)

>Forward chain - trying rule: Provides the name of a rule the forward chaining process is about to try. The >Rule status after evaluation is: trace output will follow. Is it going to rain?

No eligible rules for: The named attribute's value has been resolved but there are no eligible rules to try with forward chaining. the recommendation

REMOVE FROM GOAL STACK Reason for removing an attribute from the goal stack followed by the attribute name. Reasons are: (Resolved) meaning a value was determined or (Failed) meaning a value could not be determined. (Resolved): precipitation

REPLACE ON GOAL STACK: When a value is resolved for a multivalued attribute (MAXVALS > 1) it is removed from the goal stack. If there are any remaining unresolved values, the attribute is put back on the goal stack. the recommendation

GOAL STACK EMPTY: END CONSULTATION The goal stack is empty so the consultation is over. Each goal variable has either been resolved or the inference engine has determined that it cannot be resolved with the available facts.

Notes

Knowledge base tracing is available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. The Trace is ON/OFF button toggles the trace output on or off. This button may be used at any time during a consultation.

Knowledge Base Trace Output
Statement	Description	Examples (from various demo KBs)
STARTING NEW CONSULTATION WITH THE FOLLOWING GOALS:	List of all attributes named by GOAL statements	>the recommendation
Minimum confidence factor for accepting a value as a fact:	Provided after the list of goals when a new consultation begins.	80%
TRYING RULE:	The rule description for the rule the inference engine is about to consider.	Is it going to rain?
>Add to goal stack	Provides the name of an attribute added to the goal stack. A value of this attribute is needed to evaluate the rule the inference engine last considered leading to an Unknown status.	precipitation
>Rule status after evaluation is:	The status of the rule the inference engine just considered: possible values are Unknown, True/fired, False and Failed.	Unknown
PROMPT FOR:	A PROMPT will be tried to obtain a value for the named attribute.	precipitation
*Prompt assigned:	The value returned from the last issued PROMPT followed by the confidence with which the value was input.	precipitation=expected (100.0%)
*Rule assigned:	A rule has fired and the consequent assigned the specified value to the named attribute with the confidence level shown in parentheses.	the recommendation=wear a raincoat (100.0%)
*Default assigned:	The value specified in a knowledge base DEFAULT command was assigned to the named attribute with the confidence level shonw in parentheses after the inference engine failed to resolve its value.	the recommendation=wear whatever you want to (90.0%)
FORWARD CHAIN:	The value of an attribute has been determined from a PROMPT, because a rule fired or from a DEFAULT with sufficient confidence to be considered a fact. It will now be used to evaluate relevant rules in the knowledge base.	precipitation=expected (100.0%)
>Forward chain - trying rule:	Provides the name of a rule the forward chaining process is about to try. The >Rule status after evaluation is: trace output will follow.	Is it going to rain?
No eligible rules for:	The named attribute's value has been resolved but there are no eligible rules to try with forward chaining.	the recommendation
REMOVE FROM GOAL STACK	Reason for removing an attribute from the goal stack followed by the attribute name. Reasons are: (Resolved) meaning a value was determined or (Failed) meaning a value could not be determined.	(Resolved): precipitation
REPLACE ON GOAL STACK:	When a value is resolved for a multivalued attribute (MAXVALS > 1) it is removed from the goal stack. If there are any remaining unresolved values, the attribute is put back on the goal stack.	the recommendation
GOAL STACK EMPTY: END CONSULTATION	The goal stack is empty so the consultation is over. Each goal variable has either been resolved or the inference engine has determined that it cannot be resolved with the available facts.
Notes	Knowledge base tracing is available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. The Trace is ON/OFF button toggles the trace output on or off. This button may be used at any time during a consultation.

Knowledge Base Analysis Output

Component Description Examples

ATTRIBUTE USAGE All attribute names found in the knowledge base are listed in alphabetical order (ignoring case). PROMPTs and RULEs that can determine a value for the attribute are shown. Then, on a separate line, each RULE that uses the attribute is listed. An attribute that is used but never determined represents a logical error, and is flagged with a question mark before the attribute name. This typically occurs because an attribute name has been misspelled leading to the existence of two attributes when only one is intended. It is OK for an attribute name to be determined by never used in a rule: this will often be the case for the goal attributes. [precipitation] Determined by PROMPT 1
[precipitation] Used by rule(s) 1,2
[the recommendation] Determined by rule(s) 1,2
Error (caused by space at end of attribute name in PROMPT):
[precipitation ] Determined by PROMPT 1
?[precipitation] Used by rule(s) 1,2
[the recommendation] Determined by rule(s) 1,2

VALUE USAGE In the VALUE USAGE section of the report, all text values found in the knowledge base are listed in alphabetical order (ignoring case). Each value is followed by its length enclosed in parentheses. Then its location in the knowledge base (in a rule premise (IF), rule consequent (THEN) or PROMPT) is identified along with the associated attribute. This output helps identify values with slightly different spelling, spacing or other punctuation that makes two values that are intended to match appear unmatched to the inference engine. The length of each attribute is useful in identifying hard-to-see typographical errors such as two spaces between words when only one was intended. "expected" (8) prompt 1 value of [precipitation]
"expected" (8) rule 1 premise value of [precipitation]
"expected" (8) rule 2 premise value of [precipitation]
Error (caused by misspelling "expected" in rule 2's premise)
"expected" (8) rule 1 premise value of [precipitation]
"exprected" (9) rule 2 premise value of [precipitation]

Notes

Knowledge base analyses are available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. To obtain this output, click the Analyze KB button in the KNOWLEDGE BASE DEVELOPER'S OUTPUT window. An analysis represents static information about the knowledge base and may be requested at any time during the consultation.

Knowledge Base Analysis Output
Component	Description	Examples
ATTRIBUTE USAGE	All attribute names found in the knowledge base are listed in alphabetical order (ignoring case). PROMPTs and RULEs that can determine a value for the attribute are shown. Then, on a separate line, each RULE that uses the attribute is listed. An attribute that is used but never determined represents a logical error, and is flagged with a question mark before the attribute name. This typically occurs because an attribute name has been misspelled leading to the existence of two attributes when only one is intended. It is OK for an attribute name to be determined by never used in a rule: this will often be the case for the goal attributes.	[precipitation] Determined by PROMPT 1 [precipitation] Used by rule(s) 1,2 [the recommendation] Determined by rule(s) 1,2 Error (caused by space at end of attribute name in PROMPT): [precipitation ] Determined by PROMPT 1 ?[precipitation] Used by rule(s) 1,2 [the recommendation] Determined by rule(s) 1,2
VALUE USAGE	In the VALUE USAGE section of the report, all text values found in the knowledge base are listed in alphabetical order (ignoring case). Each value is followed by its length enclosed in parentheses. Then its location in the knowledge base (in a rule premise (IF), rule consequent (THEN) or PROMPT) is identified along with the associated attribute. This output helps identify values with slightly different spelling, spacing or other punctuation that makes two values that are intended to match appear unmatched to the inference engine. The length of each attribute is useful in identifying hard-to-see typographical errors such as two spaces between words when only one was intended.	"expected" (8) prompt 1 value of [precipitation] "expected" (8) rule 1 premise value of [precipitation] "expected" (8) rule 2 premise value of [precipitation] Error (caused by misspelling "expected" in rule 2's premise) "expected" (8) rule 1 premise value of [precipitation] "exprected" (9) rule 2 premise value of [precipitation]
Notes	Knowledge base analyses are available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. To obtain this output, click the Analyze KB button in the KNOWLEDGE BASE DEVELOPER'S OUTPUT window. An analysis represents static information about the knowledge base and may be requested at any time during the consultation.

Knowledge Base Dump Output

Component Description Examples

Minimum CF Minimum confidence factor that must be achieved before the value of an attribute is accepted as known. Minimum confidence factor for accepting a value as a fact: 80%

ATTRIBUTES The current value of an attribute followed by its type (S for string, N for numeric and B for Boolean) and the confidence currently associated with the value. If the knowledge base includes a DEFAULT statement for this attribute, the default value, type and confidence with which it will be assigned is provided. Attribute 3: the recommendation=wear your boots (S)(95%) DEFAULT: wear whatever you want to (90%)

RULES The first line for each rule provides the rule description, current status (U for unknown, T for proven true, F for proven false and X for failed (cannot be determined)) followed by the current confidence factor associated with the rule.
Next each premise (IF:) and consequent (THEN:) clause is presented along with its value type. Consequent clauses also provide the confidence factor associated with the value assignment in parentheses. RULE 1: [Is it going to rain?] Status: U (0.0%)
IF: precipitation=expected(S) and
the expected temperature>32.0(N)
THEN: the recommendation=wear a raincoat(S) (100.0%)

GOAL STACK The goal stack provides a list of the attributes the inference engine is trying to resolve. The last (highest numbered) goal is the current inferencing objective. A goal is removed when its value is determined or there is no way to determine its value. When the goal stack is empty, the consultation ends. Goal 1: the recommendation
Goal 2: the expected temperature

Notes

Knowledge base dumps are available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. To obtain this output, click the Display KB dump button in the KNOWLEDGE BASE DEVELOPER'S OUTPUT window. Dumps may be requested at any time during the consultation.
If an attribute is allowed to have multiple values (there is a MAXVALS statement for this attribute with a value greater than 1), it will appear multiple times in the ATTRIBUTE section of the dump.

Knowledge Base Dump Output
Component	Description	Examples
Minimum CF	Minimum confidence factor that must be achieved before the value of an attribute is accepted as known.	Minimum confidence factor for accepting a value as a fact: 80%
ATTRIBUTES	The current value of an attribute followed by its type (S for string, N for numeric and B for Boolean) and the confidence currently associated with the value. If the knowledge base includes a DEFAULT statement for this attribute, the default value, type and confidence with which it will be assigned is provided.	Attribute 3: the recommendation=wear your boots (S)(95%) DEFAULT: wear whatever you want to (90%)
RULES	The first line for each rule provides the rule description, current status (U for unknown, T for proven true, F for proven false and X for failed (cannot be determined)) followed by the current confidence factor associated with the rule. Next each premise (IF:) and consequent (THEN:) clause is presented along with its value type. Consequent clauses also provide the confidence factor associated with the value assignment in parentheses.	RULE 1: [Is it going to rain?] Status: U (0.0%) IF: precipitation=expected(S) and the expected temperature>32.0(N) THEN: the recommendation=wear a raincoat(S) (100.0%)
GOAL STACK	The goal stack provides a list of the attributes the inference engine is trying to resolve. The last (highest numbered) goal is the current inferencing objective. A goal is removed when its value is determined or there is no way to determine its value. When the goal stack is empty, the consultation ends.	Goal 1: the recommendation Goal 2: the expected temperature
Notes	Knowledge base dumps are available when running in the DEBUG mode obtained by setting the DEBUG parameter to TRUE in the <APPLET> parameters. To obtain this output, click the Display KB dump button in the KNOWLEDGE BASE DEVELOPER'S OUTPUT window. Dumps may be requested at any time during the consultation. If an attribute is allowed to have multiple values (there is a MAXVALS statement for this attribute with a value greater than 1), it will appear multiple times in the ATTRIBUTE section of the dump.

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