[go: up one dir, main page]
More Web Proxy on the site http://driver.im/

WO1990007152A1 - A modular blackboard-based expert system - Google Patents

A modular blackboard-based expert system Download PDF

Info

Publication number
WO1990007152A1
WO1990007152A1 PCT/US1989/005622 US8905622W WO9007152A1 WO 1990007152 A1 WO1990007152 A1 WO 1990007152A1 US 8905622 W US8905622 W US 8905622W WO 9007152 A1 WO9007152 A1 WO 9007152A1
Authority
WO
WIPO (PCT)
Prior art keywords
knowledge
rules
rule
action
eligibility
Prior art date
Application number
PCT/US1989/005622
Other languages
French (fr)
Inventor
Evangelos Simoudis
Original Assignee
Digital Equipment Corporation
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Digital Equipment Corporation filed Critical Digital Equipment Corporation
Publication of WO1990007152A1 publication Critical patent/WO1990007152A1/en

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06NCOMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00Computing arrangements using knowledge-based models
    • G06N5/04Inference or reasoning models
    • G06N5/043Distributed expert systems; Blackboards

Definitions

  • the invention relates generally to the field of rule-based expert systems.
  • Typical expert systems include a knowledge base, a working memory and a control portion, or inference engine.
  • the knowledge base contains knowledge represented in the form of either production rules or procedures or a combination of both.
  • the working memory which is also termed a "blackboard", contains a set of facts, which may be established at the beginning of operation of the system, during prior processing by the system, or by an operator.
  • the control portion controls the operation of the system in a series of execution cycles during which it processes the rules or procedures in connection ' with the contents of the working memory.
  • Production rules generally comprise two portions, namely, a condition portion, generally identified as the "left-hand side” of the rule, and an action or consequent list, which is generally identified as the "right-hand side” of the rule.
  • the condition portion identifies one or more conditions which may be satisfied by the facts in the blackboard.
  • the action list specifies one or more actions in connection with the- system, such as adding, deleting or modifying a fact on the blackboard, transmitting a message to the- operator, or halting the execution cycle. If the conditions comprising:a rule's condition portion are satisfied, the rule is said to "fire", in which case the control portion executes the operations specified in the action list.
  • Procedures are essentially computer programs that define a series of steps or operations to be performed during an execution cycle.
  • the procedure may specify, for example, one or more actions to be taken in connection with the system such as storing facts in the blackboard, transmitting a message to the operator or halting the execution cycle.
  • the invention provides a new and improved expert system.
  • the expert system includes a blackboard, a plurality of knowledge sources, a control knowledge source and a control module.
  • the blackboard stores facts used during an execution cycle.
  • Each knowledge source includes rules for performing selected operations in connection with the facts in the blackboard.
  • the control knowledge source includes selection rules for selecting among the knowledge sources.
  • the control module performing an execution cycle including an eligibility determination phase to identify one or more of the knowledge sources, a knowledge source selection phase using the selection rules in said control knowledge source to select one of the identified knowledge sources, and an action phase to process a rule of the selected- knowledge source... • ⁇ • - • .
  • Fig. 1 is a functional block diagram of an expert system constructed in accordance with the invention
  • Figs. 2A through 2C are flow diagrams useful in understanding the operation of the expert system constructed in Fig . 1 .
  • Fig. 1 depicts a functional block diagram of. an expert system constructed in accordance with the invention.
  • the system includes a blackboard 10, a plurality of knowledge sources 11A through UN (generally identified by reference numeral 11), a control knowledge rule base 12, a rule base 13, and a procedure base 14 all controlled by a control module 15.
  • the system depicted in Fig. 1 represents functional elements or computer programs processed by a computer including a processor element, memory element and input/output element under control of an operator.
  • the blackboard 10 comprises a region of the computer memory which is shared by the other functional elements of the system as described below.
  • the blackboard 10 is divided into a plurality of sub-regions, identified as levels, 16(0) through 16(N) (generally identified by reference numeral 16).
  • the rule base 13 contains one or more production rule ⁇
  • a production rule typically comprises two parts, including a condition portion and an action, or consequent, portion.
  • the condition portion which is also known as the "left-hand side" sets forth one or more conditions.
  • Each condition relates to one or more of the facts which may be present in the blackboard 10, so that, if the facts in the blackboard 10 indicate that the conditions are satisfied, r - the rule is said to "fire”. If a rule fires, the actions ; specified in the action portion are performed, as a result of which facts on the blackboard 10 may be modified, may be added to or deleted from the blackboard 10.
  • actions identified in the action portion of a rule may specify that a message be delivered to the operator through the computer's input/output element. Furthermore, the actions may specify that selected control actions, such as initialization or .termination of operation, be taken concerning the operation Of the system.
  • the procedure base 14 contains one or .more procedures each defining a sequence of operations to be- taken.
  • the operations may specify operations to be taken in connection with the facts in the blackboard 10, such as adding, deleting or modifying selected facts, may specify that messages be delivered to the operator through the computer's input/output element, and may also specify selected control actions, such as initialization or termination of operation, to be taken in connection with the operation of the system.
  • the knowledge sources 11 include information that controls selection of particular rules in the rule base 13 and/or procedures in the procedure base 14. Preliminarily, in connection with a particular application in which the system depicted in Fig. 1 may be used, each knowledge source 11 is associated with, and used in connection with, some portions or aspects of the application. In connection with the portions of the application with which the knowledge source 11 is associated, the knowledge source 11 identifies rules in the rule.base 13 and/or procedures* in the procedure base 14 that relate to the portions of the application with respect to which the knowledge source 11 is associated.
  • the rules and procedures that relate to portions of the application are termed herein "action rules" 1 and "action - procedures", respectively.
  • the knowledge source 11 identifies rules in the rule base 13 and/or procedures in the procedure base 14 that, in turn, identify the knowledge source 11 as being associated with the particular portions of the application. These rules and procedures are termed herein “eligibility rules” and “eligibility procedures”, respectively.
  • control knowledge rule base 12 contains rules that resolve conflicts among knowledge sources 11 when more than one knowledge source 11, may be associated with particular portions of an application. That is, if more: than one knowledge source 11 is associated with a portion of an application, the control knowledge rule base 12 selects one of the knowledge sources 11 to be applied.
  • the system depicted in Fig. 1 may be used to perform redesign operations in connection with an electrical circuit design based on differences between simulated operational characteristics of the circuit and desired operational characteristics of the circuit as provided by the operator.
  • the facts in the blackboard 10 comprise the identifications of the components of the electrical circuit and their interconnections, as well as the simulated operational characteristics and the desired operational characteristics.
  • the action rules and action procedures of the various knowledge sources .11 may, : for example, enable identifications of components to be deleted from or added to the blackboard 10, and may also enable interconnections among components to be changed, to correct particular discrepancies between the simulated operational characteristics and the desired operational characteristics.
  • each of the knowledge .sources 11 enable each respective, knowledge source 11 to be identified as useful in connection with a redesign operation to correct a discrepancy between the simulated. and desired operational characteristics.
  • A? ⁇ plurality of knowledge sources 11 may be useful in correcting a discrepancy, as indicated by their respective eligibility rules and procedures, and the rules in the control knowledge rule base 12 enables the selection of one particular knowledge source 11.
  • the control module 15 uses the action rules and/or procedures to actually perform a redesign operation. .
  • the control module 15 controls the other portions of the system depicted in Fig. 1 in an execution cycle, each comprising three phases. During a first phase, control module 15 identifies the knowledge " sources 11 that are.
  • control module 15 uses the rules in the rule base 13 and/or procedures in the procedure base 14 that are identified as eligibility rules and eligibility procedures, respectively, by all of the knowledge sources 11. With reference to the preceding example, during the first phase, the control module 15 identifies the knowledge sources 11 that are eligible to be used in performing a redesign operation in connection with the discrepancies between the simulated and desired operational characteristics of the electrical circuit.
  • the control module 15 After the first phase, the control module 15 enters a second phase, in which it selects one of the knowledge sources 11 using the knowledge source selection rules in the control knowledge rule base 12. After the control module 15 selects a knowledge source 11, it enters a third phase, in which it uses the action rules and/or action procedures identified by the knowledge source 11. With reference to ⁇ the preceding example, during the third phase the control module 15 performs a redesign operation enabled by action rules in the rule base 13 and/or action procedures in the procedure base 14 identified by the knowledge source 11 to end the execution cycle. . After the execution cycle,- other elements of an application system including the expert system depicted in Fig. 1 may-add, delete, or modify facts in the blackboard 10, and enable the control module to perform another execution cycle.
  • a knowledge source 11 identifies rules in the rule base 13 and/or procedures in the procedure base 14.
  • the elements in the knowledge sources 11 are similar and so only one knowledge source UN is depicted in detail .
  • the knowledge source 11 includes a set of attributes, including a source level attribute 19 and a destination level attribute 20 both of which associate the knowledge source 11 with a level ,16 in the blackboard 10.
  • the source level attribute identifies the level 16 in the blackboard 10 from which the control module 15 obtains the facts used in processing the rules and procedures associated with the knowledge source 11.
  • the action rules and action procedures may enable facts in the blackboard 10 to be added, modified or deleted, and the destination level attribute 20 identifies the level 16 in the blackboard 10 in which these operations are to occur.
  • the knowledge source 11 also includes two sets of attributes directed to the association of the rules in the rule base 13 and procedures in the procedure base 14 with the knowledge source 11, including an action attribute set 21 and an eligibility attribute set 22.
  • the action attribute set 21 includes an action rule pointer 23, an action procedure pointer 25 and an action rule strategy pointer 26.
  • the action rule pointer 23 points to an action rule pointer list 24.
  • the action rule pointer list contains pointers that point to the action rules in the rule base 13 that are associated with the knowledge source 11.
  • the action procedure pointer 25 points to an action procedure in the procedure base 14. In one.specific embodiment, the action procedure pointer 25 is empty (that is, it does not contain a pointer to an action procedure) unless the action rule pointer 23 is empty or the action -._. rule pointer list 2 ' pointed to by the action rule pointer 23 is empty.
  • the action rule strategy pointer 26 points to a set of action rules scheduling rules 27 which define the order in which the action rules identified by the action rule pointer list 24 are to be used by the control module 15 during the third phase of the execution cycle.
  • the action rules scheduling rules 27 effectively resolve conflicts among the action rules.
  • the action rules scheduling rules 27 may enable only one or some of the action rules identified by the action rule pointer, list 24 to be used, or they may enable all of them to be used in a particular order.
  • the eligibility attribute set 22 includes an eligibility rule pointer 30, an eligibility procedure pointer 32 and an eligibility rule strategy pointer 33.
  • the eligibility rule pointer 30 points to an eligibility rule pointer list 31.
  • the eligibility rule pointer list contains pointers that point to the eligibility rules in the rule base 13 that are associated with the knowledge source 11.
  • the eligibility procedure pointer 32 points to an eligibility procedure in the procedure base 14. In one specific embodiment, the eligibility procedure pointer 32 is empty (that is, it does not contain a pointer to an eligibility procedure) unless the eligibility rule pointer 30 is empty or the eligibility rule pointer list 31 pointed to by the eligibility rule pointer 30 is empty.
  • the eligibility.rule strategy pointer 33 points to a set of eligibility rules scheduling rules 34 which define the order in which the eligibility rules identified by the eligibility rule pointer list 31 are to be used by the control module 15 during the third phase o . the execution cycle.
  • the eligibility rules scheduling rules 34 effectively resolve conflicts among the eligibility rules.
  • the eligibility rules scheduling rules 34 may enable only one or some of the eligibility rules identified by the eligibility rule pointer list 31 to be used, or they may enable all of them to be used in a particular order.
  • the knowledge source 11 includes an action rule pointer list 24 and an eligibility rule pointer list 31, while the actual action rules and eligibility rules are provided in a single rule base 13, so that, if an action rule or an eligibility rule in the rule base 13 is -associated with multiple knowledge sources 11, the rule need not be duplicated among the various knowledge sources 11. Further, aggregating the action rules and eligibility rules in a common rule base simplifies maintenance and updating of the rules as may be required.
  • the scheduling rules which are shown in Fig. 1 as being included in the knowledge source 11, may also be located in the rule base 13 and pointers provided as elements 27 and 34, respectively.
  • the operations initiated by the control module 15 are in the form of three operational phases.
  • the control module 15 determines the eligibility of the various knowledge sources 11 using the eligibility rules and/or eligibility procedures defined by the eligibility attribute set 22.
  • the control module 15 uses the action rules and/or action procedures defined by the action attribute set 21 of the selected knowledge source 11.
  • the control module initially chooses a knowledge source 11 (step 100) and determines whether it has an associated eligibility rule or eligibility procedure to determine whether the knowledge source 11 is eligible to be selected during the second phase. If the eligibility rule pointer 30 and eligibility rule pointer list 31 are not empty, the knowledge source 11 has an eligibility rule, but if either the eligibility rule pointer 30 or the eligibility rule pointer list 31 are empty, the -knowledge source 11 has an associated eligibility procedure which is to be used to determine eligibility of the knowledge source 11.
  • step 101 the Control module sequences to step 102, in which it identifies, using the pointers in the eligibility rule pointer list 31 identified by the eligibility rule pointer 30, the eligibility rules in the rule base 13 which determine eligibility of the knowledge source 11.
  • the control module 15 determines whether any of the condition portions of any of the eligibility rules are satisfied by* the facts on the blackboard 10 (step 103), to determine whether any of the eligibility rules pointed to by the knowledge source 11 are in condition for firing. In this operation, the control module 15 uses the facts in the level 16 of blackboard 10 that is identified by the source level attribute 19 of the knowledge source 11. If the conditions of none of the eligibility rules are satisfied, the control module sequences to step 110.
  • the control module 15 determines whether only one eligibility rule is in condition for firing or multiple eligibility rules are eligible for firing (step 104). If only one eligibility rule is in condition for firing, the control module 15 sequences to step 106. On the other hand, if multiple eligibility rules are in condition for firing, the control module uses the eligibility rules scheduling rules 34 identified by the eligibility rule strategy pointer 33 to select one or more of the identified eligibility rules for firing (step 105).
  • control module 15 selects one or more of the eligibility rules in -li ⁇ the rule base 13 pointed to by the eligibility rule pointer list 31, it fires the selected rule(s) to identify the knowledge source 11 as being eligible (step 106).
  • the control module 15 performs the actions specified in the action list in the rule, during which it loads the identification of the knowledge source 11 on the blackboard 10 in the level identified by destination level attribute 20.
  • the knowledge source 11 specifies, in the eligibility procedure pointer 32, an eligibility procedure in procedure base 14 to determine eligibility of the knowledge source 11. Accordingly, the control module 15 processes the identified procedure, using the facts in the level 16 of blackboard 10 identified by source level attribute 19. During processing of the eligibility procedure, the control module 15 may load the identification of the knowledge source 11 on the level 16 of the blackboard 10 identified by the destination level attribute 20. After execution of the eligibility procedure, the control module 15 sequences to step 110.
  • control module 15 sequences to step 110, in which it determines whether it has determined eligibility of all of the knowledge sources 11. If not, it chooses another knowledge source 11 and returnp to step 101. On the other hand, if it has determined eligibility by all of the knowledge sources 11, it sequences to the second phase of the execution cycle, during which it selects one of the.eligible knowledge sources 11..
  • control module 15 first determines whether any of the knowledge sources 11 are identified on the blackboard 10 as being eligible (step 120). If not, the control module exits (step 121) to terminate the execution cycle. If one knowledge source 11 is identified on the blackboard 10 as being eligible (step 122), the control module 15 specifies the identified knowledge source 11 as being selected (step 123), and sequences to the third phase (step 130). Following step 122, the control module 15 determines whether the blackboard 10 identifies multiple knowledge sources 11 as eligible (step 124). If not, an error has occurred and the control module exits (step 125).
  • control module 15 determines, in step 124, that multiple knowledge sources 11 are identified on the blackboard 10 as being eligible, it uses the control knowledge rule base 12 to select one of the knowledge sources 11 identified on the blackboard 10 for use during the third phase (step 126).
  • Each rule in the control knowledge rule base 12 has a condition portion that identifies one or more of the knowledge sources 11, and an action list that enables the control module 15 to identify one of the knowledge sources 11 identified in the condition portion on the blackboard 10, to thereby select the knowledge source 11.
  • control module 15 has identified, on the blackboard 10, one of the knowledge sources 11 as being selected.
  • the control module 15 applies the knowledge source 11 that was selected during the second phase, and particularly " the action rules and/or action procedure identified by its action attribute set 21 (Fig. 1), to the facts in the blackboard 10.
  • the control module 15 first determines whether the action rule pointer 23 or action rule pointer list is empty (step 130), and, if so, it executes the action procedure in the procedure base 14 identified by the action procedure pointer 25 (step 131).
  • the control module 15 uses the facts in the level 16 of the blackboard 10 identified by the source level attribute 19 of the knowledge source 11, and may add, delete and/or modify facts in the level 16 of the blackboard 10 identified by the destination level attribute 20.
  • control module 15 may transmit a message to the operator, and/or perform control operations in connection with the system.
  • the control module 15 exits (step 132). If, in step 130, the control module 15 determines that the action rule pointer 23 and the action rule pointer list are not both empty, it sequences to step 133. In step 133, the control module 15 identifies those action rules in the rule base 13 that are pointed to by the contents of the action rule pointer list 24 that is, in turn, pointed to by the action rule pointer 23.
  • the control module 15 determines which of them are in condition for firing, that is, it determines which of the rules have condition portions that are satisfied by the facts in the level 16 in the blackboard 10 identified by the source level attribute 19.
  • the action rules may have condition portions that identify particular ones of the levels 16 in the blackboard 10, and the control module 15 determines whether such an action rule is in condition for firing based on whether the conditions correspond to the facts in the particular level 16.
  • step 134 the control module 15 determines whether none of the action rules were in condition for firing, in which case it sequences to step 134 to exit. If none of the action rules were in condition for firing, an error has occurred, since, for the eligibility rules to indicate that the knowledge source 11 is eligible, at least one of the action rules pointed to by the action rule pointer list of the knowledge source 11 should fire. If, on the other hand, the control module 15 determines in step 134 that the conditions of at least one action rule are satisfied, it sequences to step 136 to determine if the conditions of multiple action rules are satisfied.
  • step 140 the control module 15 sequences to step 140 to fire that rule. If the control module 15 determines in step 136 that the conditions of multiple action rules are satisfied, it sequences to step 137. In step 137, the control module 15 uses the action rules scheduling rules 27 identified by the action rule strategy pointer 26 to select one or more of the action rules for firing, and the order in which the action rules are to be fired if multiple action rules are to be selected. Following step 137, the control module 15 sequences to step 140 to fire the selected action rule(s), to end the execution cycle. After step 140, the control module 15 returns to step 100 to begin another execution cycle (step 141) .
  • the system depicted in Fig. 1 and 2A through 2C has a number of advantages.
  • Providing the knowledge sources 11 and control knowledge rule base 12 separate and apart from the rule base 13 and the control module 15 helps to modularize the system in accordance with the application in which the system is used. Since the diverse knowledge sources 11 are associated with different portions of an application, updating and maintenance of the system are simplified, since only the knowledge sources associated with the portions of the updated system need be modified or updated. Furthermore, dividing the blackboard 10 into levels 16 helps to modularize the facts as used in connection with the rules.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

An expert system includes a blackboard, a plurality of knowledge sources, a control knowledge source and a control module. The blackboard stores facts used during an execution cycle. Each knowledge source includes rules for performing selected operations in connection with the facts in the blackboard. The control knowledge source includes selection rules for selecting among the knowledge sources. The control module performing an execution cycle including an eligibility determination phase to identify one or more of the knowledge sources, a knowledge source selection phase using the selection rules in said control knowledge source to select one of the identified knowledge sources, and an action phase to process a rule of the selected knowledge source.

Description

A MODULAR BLACKBOARD-BASED EXPERT SYSTEM
FIELD OF THE INVENTION
The invention relates generally to the field of rule-based expert systems.
BACKGROUND OF THE INVENTION
Typical expert systems include a knowledge base, a working memory and a control portion, or inference engine. The knowledge base contains knowledge represented in the form of either production rules or procedures or a combination of both. The working memory, which is also termed a "blackboard", contains a set of facts, which may be established at the beginning of operation of the system, during prior processing by the system, or by an operator. The control portion controls the operation of the system in a series of execution cycles during which it processes the rules or procedures in connection' with the contents of the working memory.
Production rules generally comprise two portions, namely, a condition portion, generally identified as the "left-hand side" of the rule, and an action or consequent list, which is generally identified as the "right-hand side" of the rule. The condition portion identifies one or more conditions which may be satisfied by the facts in the blackboard. The action list specifies one or more actions in connection with the- system, such as adding, deleting or modifying a fact on the blackboard, transmitting a message to the- operator, or halting the execution cycle. If the conditions comprising:a rule's condition portion are satisfied, the rule is said to "fire", in which case the control portion executes the operations specified in the action list.
Procedures are essentially computer programs that define a series of steps or operations to be performed during an execution cycle. The procedure may specify, for example, one or more actions to be taken in connection with the system such as storing facts in the blackboard, transmitting a message to the operator or halting the execution cycle.
SUMMARY OF THE INVENTION
The invention provides a new and improved expert system.
In brief summary, the expert system includes a blackboard, a plurality of knowledge sources, a control knowledge source and a control module. ϊThe blackboard stores facts used during an execution cycle. Each knowledge source includes rules for performing selected operations in connection with the facts in the blackboard. The control knowledge source includes selection rules for selecting among the knowledge sources. The control module performing an execution cycle including an eligibility determination phase to identify one or more of the knowledge sources, a knowledge source selection phase using the selection rules in said control knowledge source to select one of the identified knowledge sources, and an action phase to process a rule of the selected- knowledge source... • • -.
BRIEF DESCRIPTION OF THE DRAWINGS
This invention is pointed out with particularity in the appended claims. The above and further advantages of this invention may be better understood by referring to the following description- taken in conjunction with the accompanying drawings, in which:
Fig. 1 is a functional block diagram of an expert system constructed in accordance with the invention;
Figs. 2A through 2C are flow diagrams useful in understanding the operation of the expert system constructed in Fig . 1 .
DETAILED DESCRIPTION OF AN ILLUSTRATIVE EMBODIMENT
Fig. 1 depicts a functional block diagram of. an expert system constructed in accordance with the invention. With reference to Fig. 1, the system includes a blackboard 10, a plurality of knowledge sources 11A through UN (generally identified by reference numeral 11), a control knowledge rule base 12, a rule base 13, and a procedure base 14 all controlled by a control module 15. As is conventional, the system depicted in Fig. 1 represents functional elements or computer programs processed by a computer including a processor element, memory element and input/output element under control of an operator.
The blackboard 10 comprises a region of the computer memory which is shared by the other functional elements of the system as described below. The blackboard 10 is divided into a plurality of sub-regions, identified as levels, 16(0) through 16(N) (generally identified by reference numeral 16).
The rule base 13 -contains one or more production ruleε A production rule typically comprises two parts, including a condition portion and an action, or consequent, portion. Briefly, the condition portion, which is also known as the "left-hand side", sets forth one or more conditions. Each condition relates to one or more of the facts which may be present in the blackboard 10, so that, if the facts in the blackboard 10 indicate that the conditions are satisfied, r- the rule is said to "fire". If a rule fires, the actions ; specified in the action portion are performed, as a result of which facts on the blackboard 10 may be modified, may be added to or deleted from the blackboard 10. Additionally, actions identified in the action portion of a rule may specify that a message be delivered to the operator through the computer's input/output element. Furthermore, the actions may specify that selected control actions, such as initialization or .termination of operation, be taken concerning the operation Of the system.
The procedure base 14 contains one or .more procedures each defining a sequence of operations to be- taken. The operations may specify operations to be taken in connection with the facts in the blackboard 10, such as adding, deleting or modifying selected facts, may specify that messages be delivered to the operator through the computer's input/output element, and may also specify selected control actions, such as initialization or termination of operation, to be taken in connection with the operation of the system.
* The knowledge sources 11 include information that controls selection of particular rules in the rule base 13 and/or procedures in the procedure base 14. Preliminarily, in connection with a particular application in which the system depicted in Fig. 1 may be used, each knowledge source 11 is associated with, and used in connection with, some portions or aspects of the application. In connection with the portions of the application with which the knowledge source 11 is associated, the knowledge source 11 identifies rules in the rule.base 13 and/or procedures* in the procedure base 14 that relate to the portions of the application with respect to which the knowledge source 11 is associated. The rules and procedures that relate to portions of the application are termed herein "action rules"1 and "action - procedures", respectively.
In addition, the knowledge source 11 identifies rules in the rule base 13 and/or procedures in the procedure base 14 that, in turn, identify the knowledge source 11 as being associated with the particular portions of the application. These rules and procedures are termed herein "eligibility rules" and "eligibility procedures", respectively.
Finally, the control knowledge rule base 12 contains rules that resolve conflicts among knowledge sources 11 when more than one knowledge source 11, may be associated with particular portions of an application. That is, if more: than one knowledge source 11 is associated with a portion of an application, the control knowledge rule base 12 selects one of the knowledge sources 11 to be applied.
For example, in one embodiment, the system depicted in Fig. 1 may be used to perform redesign operations in connection with an electrical circuit design based on differences between simulated operational characteristics of the circuit and desired operational characteristics of the circuit as provided by the operator. In that embodiment, the facts in the blackboard 10 comprise the identifications of the components of the electrical circuit and their interconnections, as well as the simulated operational characteristics and the desired operational characteristics. The action rules and action procedures of the various knowledge sources .11 may,: for example, enable identifications of components to be deleted from or added to the blackboard 10, and may also enable interconnections among components to be changed, to correct particular discrepancies between the simulated operational characteristics and the desired operational characteristics.
On the other hand, the eligibility rules and procedures of each of the knowledge .sources 11 enable each respective, knowledge source 11 to be identified as useful in connection with a redesign operation to correct a discrepancy between the simulated. and desired operational characteristics. A?< plurality of knowledge sources 11 may be useful in correcting a discrepancy, as indicated by their respective eligibility rules and procedures, and the rules in the control knowledge rule base 12 enables the selection of one particular knowledge source 11. After the control module 15 selects a knowledge source 11, it uses the action rules and/or procedures to actually perform a redesign operation. .The control module 15 controls the other portions of the system depicted in Fig. 1 in an execution cycle, each comprising three phases. During a first phase, control module 15 identifies the knowledge" sources 11 that are. associated with the particular portions of the application. In this operation, the control module 15 uses the rules in the rule base 13 and/or procedures in the procedure base 14 that are identified as eligibility rules and eligibility procedures, respectively, by all of the knowledge sources 11. With reference to the preceding example, during the first phase, the control module 15 identifies the knowledge sources 11 that are eligible to be used in performing a redesign operation in connection with the discrepancies between the simulated and desired operational characteristics of the electrical circuit.
*
After the first phase, the control module 15 enters a second phase, in which it selects one of the knowledge sources 11 using the knowledge source selection rules in the control knowledge rule base 12. After the control module 15 selects a knowledge source 11, it enters a third phase, in which it uses the action rules and/or action procedures identified by the knowledge source 11. With reference to the preceding example, during the third phase the control module 15 performs a redesign operation enabled by action rules in the rule base 13 and/or action procedures in the procedure base 14 identified by the knowledge source 11 to end the execution cycle. . After the execution cycle,- other elements of an application system including the expert system depicted in Fig. 1 may-add, delete, or modify facts in the blackboard 10, and enable the control module to perform another execution cycle.
*? ""
As described above, a knowledge source 11 identifies rules in the rule base 13 and/or procedures in the procedure base 14. The elements in the knowledge sources 11 are similar and so only one knowledge source UN is depicted in detail .
The knowledge source 11 includes a set of attributes, including a source level attribute 19 and a destination level attribute 20 both of which associate the knowledge source 11 with a level ,16 in the blackboard 10. The source level attribute identifies the level 16 in the blackboard 10 from which the control module 15 obtains the facts used in processing the rules and procedures associated with the knowledge source 11. In addition, as noted above, the action rules and action procedures may enable facts in the blackboard 10 to be added, modified or deleted, and the destination level attribute 20 identifies the level 16 in the blackboard 10 in which these operations are to occur.
The knowledge source 11 also includes two sets of attributes directed to the association of the rules in the rule base 13 and procedures in the procedure base 14 with the knowledge source 11, including an action attribute set 21 and an eligibility attribute set 22. The action attribute set 21 includes an action rule pointer 23, an action procedure pointer 25 and an action rule strategy pointer 26. The action rule pointer 23 points to an action rule pointer list 24. The action rule pointer list, in turn, contains pointers that point to the action rules in the rule base 13 that are associated with the knowledge source 11. The action procedure pointer 25 points to an action procedure in the procedure base 14. In one.specific embodiment, the action procedure pointer 25 is empty (that is, it does not contain a pointer to an action procedure) unless the action rule pointer 23 is empty or the action -._. rule pointer list 2 ' pointed to by the action rule pointer 23 is empty.
Finally, the action rule strategy pointer 26 points to a set of action rules scheduling rules 27 which define the order in which the action rules identified by the action rule pointer list 24 are to be used by the control module 15 during the third phase of the execution cycle. The action rules scheduling rules 27 effectively resolve conflicts among the action rules. Depending on the application, the action rules scheduling rules 27 may enable only one or some of the action rules identified by the action rule pointer, list 24 to be used, or they may enable all of them to be used in a particular order.
Similarly, the eligibility attribute set 22 includes an eligibility rule pointer 30, an eligibility procedure pointer 32 and an eligibility rule strategy pointer 33. The eligibility rule pointer 30. points to an eligibility rule pointer list 31. The eligibility rule pointer list, in turn, contains pointers that point to the eligibility rules in the rule base 13 that are associated with the knowledge source 11. The eligibility procedure pointer 32 points to an eligibility procedure in the procedure base 14. In one specific embodiment, the eligibility procedure pointer 32 is empty (that is, it does not contain a pointer to an eligibility procedure) unless the eligibility rule pointer 30 is empty or the eligibility rule pointer list 31 pointed to by the eligibility rule pointer 30 is empty.
Finally, the eligibility.rule strategy pointer 33 points to a set of eligibility rules scheduling rules 34 which define the order in which the eligibility rules identified by the eligibility rule pointer list 31 are to be used by the control module 15 during the third phase o . the execution cycle. , The eligibility rules scheduling rules 34 effectively resolve conflicts among the eligibility rules. Depending on the application, the eligibility rules scheduling rules 34 may enable only one or some of the eligibility rules identified by the eligibility rule pointer list 31 to be used, or they may enable all of them to be used in a particular order. The knowledge source 11 includes an action rule pointer list 24 and an eligibility rule pointer list 31, while the actual action rules and eligibility rules are provided in a single rule base 13, so that, if an action rule or an eligibility rule in the rule base 13 is -associated with multiple knowledge sources 11, the rule need not be duplicated among the various knowledge sources 11. Further, aggregating the action rules and eligibility rules in a common rule base simplifies maintenance and updating of the rules as may be required. The scheduling rules, which are shown in Fig. 1 as being included in the knowledge source 11, may also be located in the rule base 13 and pointers provided as elements 27 and 34, respectively.
With this background, the operations initiated by the control module 15 in connection with the system depicted in Fig. 1 will be described in detail in connection with Figs. 2A through 2C. As described above, the operations initiated by the control module 15 are in the form of three operational phases. In the first phase, comprising steps 100 through 107 and 110 through 111 (Fig. 2A) , the control module 15 determines the eligibility of the various knowledge sources 11 using the eligibility rules and/or eligibility procedures defined by the eligibility attribute set 22. In the second phase, comprising steps 120 through 126 (Fig. 2B), the control module 15, using the rules in the control knowledge rule base 12, selects one of the knowledge sources 11. Finally, in the third phase, comprising steps 130 through 137 and 140, the control module 15 uses the action rules and/or action procedures defined by the action attribute set 21 of the selected knowledge source 11.
With reference to Fig. 2A, in the first phase the control module initially chooses a knowledge source 11 (step 100) and determines whether it has an associated eligibility rule or eligibility procedure to determine whether the knowledge source 11 is eligible to be selected during the second phase. If the eligibility rule pointer 30 and eligibility rule pointer list 31 are not empty, the knowledge source 11 has an eligibility rule, but if either the eligibility rule pointer 30 or the eligibility rule pointer list 31 are empty, the -knowledge source 11 has an associated eligibility procedure which is to be used to determine eligibility of the knowledge source 11.
Thus, if neithef the eligibility rule pointer 30 nor the eligibility rule pointer list 31 are empty (step 101), the Control module sequences to step 102, in which it identifies, using the pointers in the eligibility rule pointer list 31 identified by the eligibility rule pointer 30, the eligibility rules in the rule base 13 which determine eligibility of the knowledge source 11. The control module 15 then determines whether any of the condition portions of any of the eligibility rules are satisfied by* the facts on the blackboard 10 (step 103), to determine whether any of the eligibility rules pointed to by the knowledge source 11 are in condition for firing. In this operation, the control module 15 uses the facts in the level 16 of blackboard 10 that is identified by the source level attribute 19 of the knowledge source 11. If the conditions of none of the eligibility rules are satisfied, the control module sequences to step 110.
If any of the eligibility rules pointed to by the knowledge source 11 are in condition for firing, that is, if any of their condition portions are satisfied by the facts in the blackboard 10, the control module 15 then determines whether only one eligibility rule is in condition for firing or multiple eligibility rules are eligible for firing (step 104). If only one eligibility rule is in condition for firing, the control module 15 sequences to step 106. On the other hand, if multiple eligibility rules are in condition for firing, the control module uses the eligibility rules scheduling rules 34 identified by the eligibility rule strategy pointer 33 to select one or more of the identified eligibility rules for firing (step 105).
Following steps 104 or 105, that is, after the control module 15 selects one or more of the eligibility rules in -li¬ the rule base 13 pointed to by the eligibility rule pointer list 31, it fires the selected rule(s) to identify the knowledge source 11 as being eligible (step 106). In that operation, for each selected rule, the control module 15 performs the actions specified in the action list in the rule, during which it loads the identification of the knowledge source 11 on the blackboard 10 in the level identified by destination level attribute 20.
Returning to step 101, if the control module 15 determines that either the eligibility rule pointer 30 or the eligibility rule pointer list 31 is empty (step 101), the knowledge source 11 specifies, in the eligibility procedure pointer 32, an eligibility procedure in procedure base 14 to determine eligibility of the knowledge source 11. Accordingly, the control module 15 processes the identified procedure, using the facts in the level 16 of blackboard 10 identified by source level attribute 19. During processing of the eligibility procedure, the control module 15 may load the identification of the knowledge source 11 on the level 16 of the blackboard 10 identified by the destination level attribute 20. After execution of the eligibility procedure, the control module 15 sequences to step 110.
Following steps 103, 106 or 107, the control module 15 sequences to step 110, in which it determines whether it has determined eligibility of all of the knowledge sources 11. If not, it chooses another knowledge source 11 and returnp to step 101. On the other hand, if it has determined eligibility by all of the knowledge sources 11, it sequences to the second phase of the execution cycle, during which it selects one of the.eligible knowledge sources 11..
More specifically, during the second phase the control module 15 first determines whether any of the knowledge sources 11 are identified on the blackboard 10 as being eligible (step 120). If not, the control module exits (step 121) to terminate the execution cycle. If one knowledge source 11 is identified on the blackboard 10 as being eligible (step 122), the control module 15 specifies the identified knowledge source 11 as being selected (step 123), and sequences to the third phase (step 130). Following step 122, the control module 15 determines whether the blackboard 10 identifies multiple knowledge sources 11 as eligible (step 124). If not, an error has occurred and the control module exits (step 125).
However, if the control module 15 determines, in step 124, that multiple knowledge sources 11 are identified on the blackboard 10 as being eligible, it uses the control knowledge rule base 12 to select one of the knowledge sources 11 identified on the blackboard 10 for use during the third phase (step 126). Each rule in the control knowledge rule base 12 has a condition portion that identifies one or more of the knowledge sources 11, and an action list that enables the control module 15 to identify one of the knowledge sources 11 identified in the condition portion on the blackboard 10, to thereby select the knowledge source 11.
Thus, at the end of the second phase, either following step 123 or step 126, the control module 15 has identified, on the blackboard 10, one of the knowledge sources 11 as being selected.
During the third phase, the control module 15 applies the knowledge source 11 that was selected during the second phase, and particularly"the action rules and/or action procedure identified by its action attribute set 21 (Fig. 1), to the facts in the blackboard 10. In the third phase, the control module 15 first determines whether the action rule pointer 23 or action rule pointer list is empty (step 130), and, if so, it executes the action procedure in the procedure base 14 identified by the action procedure pointer 25 (step 131). In executing the action procedure, the control module 15 uses the facts in the level 16 of the blackboard 10 identified by the source level attribute 19 of the knowledge source 11, and may add, delete and/or modify facts in the level 16 of the blackboard 10 identified by the destination level attribute 20. In addition, or alternatively, during processing of the action procedure the control module 15 may transmit a message to the operator, and/or perform control operations in connection with the system. After execution of the action procedure, the control module 15 exits (step 132). If, in step 130, the control module 15 determines that the action rule pointer 23 and the action rule pointer list are not both empty, it sequences to step 133. In step 133, the control module 15 identifies those action rules in the rule base 13 that are pointed to by the contents of the action rule pointer list 24 that is, in turn, pointed to by the action rule pointer 23. After identifying the action rules, the control module 15 determines which of them are in condition for firing, that is, it determines which of the rules have condition portions that are satisfied by the facts in the level 16 in the blackboard 10 identified by the source level attribute 19. In particular, the action rules may have condition portions that identify particular ones of the levels 16 in the blackboard 10, and the control module 15 determines whether such an action rule is in condition for firing based on whether the conditions correspond to the facts in the particular level 16.
After identifying the action rules in condition for firing, the control module 15 sequences to. step 134. In step 134, the control module 15 determines whether none of the action rules were in condition for firing, in which case it sequences to step 134 to exit. If none of the action rules were in condition for firing, an error has occurred, since, for the eligibility rules to indicate that the knowledge source 11 is eligible, at least one of the action rules pointed to by the action rule pointer list of the knowledge source 11 should fire. If, on the other hand, the control module 15 determines in step 134 that the conditions of at least one action rule are satisfied, it sequences to step 136 to determine if the conditions of multiple action rules are satisfied. If not, the conditions of only a single action rule are satisfied, and the control module 15 sequences to step 140 to fire that rule. If the control module 15 determines in step 136 that the conditions of multiple action rules are satisfied, it sequences to step 137. In step 137, the control module 15 uses the action rules scheduling rules 27 identified by the action rule strategy pointer 26 to select one or more of the action rules for firing, and the order in which the action rules are to be fired if multiple action rules are to be selected. Following step 137, the control module 15 sequences to step 140 to fire the selected action rule(s), to end the execution cycle. After step 140, the control module 15 returns to step 100 to begin another execution cycle (step 141) .
The system depicted in Fig. 1 and 2A through 2C has a number of advantages. Providing the knowledge sources 11 and control knowledge rule base 12 separate and apart from the rule base 13 and the control module 15 helps to modularize the system in accordance with the application in which the system is used. Since the diverse knowledge sources 11 are associated with different portions of an application, updating and maintenance of the system are simplified, since only the knowledge sources associated with the portions of the updated system need be modified or updated. Furthermore, dividing the blackboard 10 into levels 16 helps to modularize the facts as used in connection with the rules.
The foregoing description has been limited to a specific embodiment of this invention. It will be apparent, however, that variations and modifications may be made to the invention, with the attainment of some or all of the advantages of the invention. Therefore, it is the object of the appended claims to cover all such variations and modifications as come within the true spirit and scope of the invention.

Claims

1. An expert system comprising:
A. a blackboard for storing facts;
B. a plurality of knowledge sources including means defining selected operations to be performed in connection with said facts in said blackboard;
C. a control knowledge source including means defining selection among said knowledge sources; and
D. control means for performing an execution cycle including an eligibility determination phase to identify one or more of the knowledge sources, a knowledge source selection phase using said selection means in said control knowledge source to select one of said identified knowledge sources, and an action phase to process an operation defining means of said selected knowledge source.
PCT/US1989/005622 1988-12-14 1989-12-11 A modular blackboard-based expert system WO1990007152A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US28419888A 1988-12-14 1988-12-14
US284,198 1988-12-14

Publications (1)

Publication Number Publication Date
WO1990007152A1 true WO1990007152A1 (en) 1990-06-28

Family

ID=23089260

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US1989/005622 WO1990007152A1 (en) 1988-12-14 1989-12-11 A modular blackboard-based expert system

Country Status (3)

Country Link
EP (1) EP0448641A1 (en)
JP (1) JPH04503125A (en)
WO (1) WO1990007152A1 (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0570513A1 (en) * 1991-02-05 1993-11-24 Storage Technology Corporation Hierarchical distributed knowledge based machine initiated maintenance system
WO1996039658A1 (en) * 1995-06-05 1996-12-12 Westinghouse Electric Corporation Integrated information system for an industrial process and an external entity

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
AMERICAN ASSOCIATION FOR ARTIFICIAL INTELLIGENCE (AAAI), Vol. 86, pages 904-909, B. HAYES-ROTH et al., "Protean: Deriving Protein Structure from Constraints". *
ARTIFICIAL INTELLIGENCE, Vol. 26, 1985, ELSEVIER SCI. PUBL. BV, pages 251-321, B. HAYES-ROTH, "A Blackboard Architecture for Control". *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0570513A1 (en) * 1991-02-05 1993-11-24 Storage Technology Corporation Hierarchical distributed knowledge based machine initiated maintenance system
EP0570513A4 (en) * 1991-02-05 1996-06-26 Storage Technology Corp Hierarchical distributed knowledge based machine initiated maintenance system
WO1996039658A1 (en) * 1995-06-05 1996-12-12 Westinghouse Electric Corporation Integrated information system for an industrial process and an external entity

Also Published As

Publication number Publication date
JPH04503125A (en) 1992-06-04
EP0448641A1 (en) 1991-10-02

Similar Documents

Publication Publication Date Title
US5101362A (en) Modular blackboard based expert system
US5016204A (en) Expert system for performing diagnostic and redesign operations incorporating multiple levels of simulation detail
JP4117920B2 (en) How to determine core loading configuration
Crowley-Milling et al. The NODAL System for the SPS
US5291427A (en) Method for assisting the development of a set of communicating automata
US5339420A (en) Partitioning case statements for optimal execution performance
US5208768A (en) Expert system including arrangement for acquiring redesign knowledge
CN110488738A (en) A kind of code generating method and device
US5072405A (en) RETE network with provisional satisfaction of node conditions
JPH0752420B2 (en) I / O device address method
WO1990007152A1 (en) A modular blackboard-based expert system
US6014514A (en) System for generating and graphically displaying call stack information for processing elements in a parallel processing system
EP0318984B1 (en) Interrogating device
EP0367377A2 (en) System and method for producing discrimination nets for expert systems
US5604842A (en) Fuzzy reasoning processor and method, and rule setting apparatus and method
JPH0764766A (en) Maximum and minimum value calculating method for parallel computer
US5023807A (en) System and method for producing discrimination nets for expert systems
Bracchi et al. An interactive software system for computer-aided design: an application to circuit project
Katz et al. An experiment in non-procedural programming
JP3036900B2 (en) Assembler processing unit
JP3047976B2 (en) Function processing method
JPS63291128A (en) Production system
JPH05257703A (en) Assembler device
JPS6362038A (en) Control system for inference procedure
Barta PACOL: A parallel control language

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): JP

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): AT BE CH DE ES FR GB IT LU NL SE

WWE Wipo information: entry into national phase

Ref document number: 1990901395

Country of ref document: EP

WWP Wipo information: published in national office

Ref document number: 1990901395

Country of ref document: EP

WWW Wipo information: withdrawn in national office

Ref document number: 1990901395

Country of ref document: EP