JPH07271588A - Knowledge input method for example base inference - Google Patents

Abstract

PURPOSE:To input expert knowledge in an expert system by the operator at the job site and to improve the development efficiency and quality of the expert system using example base inference by using the editor screen of a general chart form for inputting knowledge. CONSTITUTION:First, the knowledge base registration 2-1 of a knowledge name and a version, etc., is performed for respective problems to be objects. Then, the input and editing 2-2 of static knowledge (a fact, a constant and external function definition) and example knowledge are performed. They are used in dynamic knowledge and dynamic example knowledge (the rules of forms that are 'If..., ... is done' and 'If the example of ... is present, ... is done'). Then, the input and editing 2-3 of inference control knowledge (meta rules) for deciding the structure and inference method of the dynamic knowledge and the dynamic example knowledge are performed. Finally, the input and editing 2-4 of the dynamic knowledge and the dynamic example knowledge are performed and the expert system is constructed. Since the editor screen of the chart form is filled by words used daily by a user, grammar is simplified and facilitated and the need of the knowledge of a high level relating to AI is eliminated.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention is a knowledge representation method for constructing an expert system in which it is effective to use a large amount of past design cases and judgment cases such as quality design and order acceptance judgment in the steel industry. , The knowledge input editing method. This is useful when a field operator who does not know the specialized knowledge of a computer wants to input his or her own specialized knowledge into the expert system, or when it is desired to improve the development efficiency, quality, and maintainability of the expert system.

[0002]

2. Description of the Related Art There is already a technique for enabling knowledge input of a forward reasoning expert system by a tabular editor (Japanese Patent Laid-Open No. 04-190425).
It systematizes the knowledge representation method of the expert system into four types of inference control knowledge, dynamic knowledge, static knowledge, and uncertain knowledge, and tabular editor screens (rules, This is a technology for inputting and editing knowledge using meta-rules, frames, facts, fuzzy, constants, data with certainty, external functions).

[0003]

On the other hand, there is a technique (Japanese Patent Laid-Open No. 03-132826) that applies case-based reasoning for solving a current problem based on past problem-solving cases. There is no technology that enables knowledge input of case-based reasoning with a tabular editor,
When inputting knowledge in case-based reasoning, knowledge is usually input in a free format in AI language.

Conventionally, when constructing an expert system using case-based reasoning, the level of knowledge representation is low,
It lacked flexibility and had problems with the development efficiency, quality, and maintainability of the expert system. Moreover, the grammar of the AI language is complicated and requires a high level of expertise on AI.
It was difficult for a person other than a computer expert to use it effectively, and there were problems that system development and changes could not be handled efficiently and could not be handled in detail.

The present invention improves the conventional knowledge input method of the expert system so that the knowledge input of case-based reasoning can also be performed so that the operator in the field who does not know the specialized knowledge of the computer can convert the specialized knowledge into the expert system. The purpose is to improve the development efficiency, quality, and maintainability of the expert system using case-based reasoning by enabling input.

[0006]

According to the present invention, a knowledge representation method for an expert system using case-based reasoning is provided.
Inference control knowledge, dynamic knowledge, static knowledge, case knowledge that is a set of past problem solving cases, and case knowledge that is highly similar to the problem to be solved at present are searched and corrected according to the degree of similarity, and the result is Systematize into 5 types of dynamic case knowledge, which is stored as new problem solving cases, and
A general-purpose tabular editor screen for inputting knowledge is displayed, and the information input to the fields displayed on the editor screen is read and edited.

[0007]

[Action] That is, the knowledge representation method of the expert system using case-based reasoning is attempted to solve inference control knowledge, dynamic knowledge, static knowledge, case knowledge which is a set of past problem solving cases, and present knowledge Dynamic case knowledge, which retrieves case knowledge that is highly similar to the problem, corrects it according to the degree of similarity, and stores the result as a new problem solving case, and inputs the knowledge about each.

As the inference control knowledge, a rule execution control procedure is input / edited on the metarule editor screen.
For dynamic knowledge, enter / edit rules (rules) regarding how to operate static knowledge from the rule editor screen. As static knowledge, from the fact editor screen,
Enter declarative knowledge to describe facts, work data, etc. declarative knowledge to describe constants from the constant editor screen, and enter / edit the definition of the external function used in the rule from the external function editor screen. As the dynamic case knowledge, a rule editor screen is used to input / edit rules (rules) regarding retrieval, correction, and storage of case knowledge consisting of a set of past problem solving cases. As case knowledge, declarative knowledge describing facts, work data, etc. used as cases is entered and edited from the case definition editor screen.

Furthermore, since the user fills the tabular editor screen with the words (Japanese) that the user uses everyday, the grammar becomes simple and easy, and no advanced knowledge about AI is required. The editor screen operates in a multi-window, mouse, and color graphics environment,
Editing work can be done efficiently. Therefore, the expression level of knowledge has increased, and knowledge input and editing have become easy.

[0010]

EXAMPLES Next, examples of the present invention will be described. In this embodiment, the computer for expert system development
The extraction pitch of the slab from the heating furnace of the hot rolling line is determined by using case-based reasoning. FIG. 1 is a block diagram showing an expert system and a development system for controlling equipment for implementing the present invention. Operation and processing parameters of the line are controlled by an equipment control computer 1-4, and the equipment control computer 1-
4 receives control data from the upper process computer 1-2. The extraction pitch determined by the expert system development computer 1-1 is set in the process computer 1-2, and the process computer 1-2 generates control data on the basis of the extracted pitch, and the equipment control computer 1-4 is generated. Give to. Alternatively, the expert system online execution process computer 1-3 is connected to the upper process computer 1-2 of the equipment control computer 1-4, and the expert system online execution process computer 1-3 is connected to the expert system online execution process computer 1-3. The extraction pitch of the slab determined by the computer 1-1 is set, and based on this, the expert system online execution process computer 1-3
Generates the control data and gives it to the process computer 1-2, which is the host computer of the facility control computer.

The knowledge input / representation method of the expert system using case-based reasoning, which will be described below, is implemented in the expert system development computer 0-1. When constructing an expert system using case-based reasoning, not only case-based reasoning is used, but also forward reasoning is usually used together to construct the system. The parts (facts, constants, external functions, metarules) relating to forward inference in this embodiment are the same as those disclosed in Japanese Patent Laid-Open No. 04-190425.

FIG. 2 shows an outline of processing in the expert system developing computer 1-1. First, the knowledge base registration 2-1 such as the knowledge name and version is performed for each target problem. Next, input and edit 2-2 of static knowledge (fact, constant, external function definition) and case knowledge is performed. These are used in dynamic knowledge and dynamic case knowledge (rules of the form: if, what to do, if there is an example to do, what to do). Next, the input editing 2-3 of the inference control knowledge (metarule) that determines the dynamic knowledge, the structure of the dynamic case knowledge, and the inference method is performed. Finally, the dynamic knowledge and dynamic case knowledge (rules) are input and edited 2-4 to construct an expert system. It is not necessary to define the static knowledge that is not used in the dynamic knowledge and the dynamic case knowledge.

FIG. 3 is a flowchart showing an online execution procedure of the case-based reasoning application expert system in the expert system online execution process computer 1-3. The case knowledge 3-1 stores a set of past problem solving cases, and the preprocessing 3-2 described by the dynamic knowledge sets the operating conditions for the case-based reasoning. Next, in case search 3-3, after setting search conditions for the current problem, a set of past problem solution cases is searched, and in case correction 3-4, the current problem and the set of past problem solution cases retrieved are searched. Is calculated, and the setting result obtained by case-based reasoning in the past problem solving case is corrected by this similarity and set as the setting value of the current problem. In the case storage 3-5, the result and the operating conditions of the current problem are stored in the case knowledge 3-1 as a problem solution case so that it can be used for the next case-based reasoning. Finally, in the post-processing 3-6 described by the dynamic knowledge, the set value obtained this time is transmitted to the process computer 1-2. The above 3-3,3-
A series of processes 4 and 3-5 is generally called case-based reasoning.

Next, detailed input and editing methods of knowledge at each stage will be described. When the expert system construction software of the computer 1-1 is started, the entire system management screen shown in FIG. 4 is displayed on the CRT. When the operator designates the item shown in it, the detailed screen of the designated item shown in FIG. 5, FIG. 6 or FIG. 7 is displayed. That is, by specifying items on the overall system management screen shown in FIG. 4, registration of a knowledge base (“Knowledge base edit” in the figure) and knowledge input / edit (“Editor” in the figure) described below are performed. In addition to selecting and starting, it is possible to select and start AI language conversion software, verify the constructed expert system, and test functions.

Knowledge base registration: 2-1 Overall system management screen of FIG. 2 When the knowledge base edit screen of FIG. 4 is selected, the knowledge base edit screen shown in FIG. 8 is displayed; "Knowledge base registration" on this screen is selected. However, the knowledge base name (“test1” in this example), version number. (The number of old and new knowledge is expressed by numbers. In this example, 1.0) and global group number (file numbers for passing inference results etc. online with the process computer are assigned numbers for debugging and programming. The two for the actual machine can be specified in this embodiment) and registered. After registration, select Finish,
Close this screen.

Input and edit of static knowledge and case knowledge: 2-2 in FIG. 2 As a method of expressing static knowledge and case knowledge, (1) case,
(2) Fact, (3) Constant, (4) 4 of external function definition
Use kind. These knowledge are the dynamic knowledge, which will be described later,
It defines the name of knowledge to be referred to or rewritten by dynamic case knowledge (rule), its initial value, and the like.

(1) Case: This editor is newly set in case-based inference, and defines a set of facts (case facts) used as cases. A case fact is a representation of various knowledge phenomena used as a case in the form of "... The fact name, field name, value, etc. defined as an example can be found in the conditional part "if ~ ~ ~
Is referred to for the case retrieved by ", and is rewritten (corrected, stored) by the rule execution unit's" to perform ~ "and used as data for controlling the case. . When the "Case" item on the screen of FIG. 5 is designated, the case definition editor screen shown in FIG. 12 is displayed. Input on this screen. This screen is: 1) Classification of functional parts for inputting knowledge data in tabular form, case facts, case fields, number, attributes, values,
Memo field 2) Functional portion for selecting or instructing the input method and editing method for each item 3) A portion indicated by 1 in FIG. 12 3) Functional portion for selecting or instructing a page of a screen to be input or displayed 4) Function part message field for displaying a system message to the user 5) After inputting data, it is composed of a function part check field that specifies whether or not there is a rationality check such as grammatical error, and knowledge data consists of the following items. input.

Classification The user groups the data and manually inputs a number to improve the visibility of the screen. Can be omitted; case fact name It is the name of the knowledge "~ no", which the user arbitrarily inputs; Case field name It is the item of knowledge "~ no ~ ga", which is used in combination with the fact name ; Case field element An array can be specified in the case field name, and the element number of the array is specified at that time; Case field array number Enter the number of elements in the array.

Input the attribute of the attribute value (“integer”, “real number”, “character string” or “integer file”, “real number file”, “character string file”); Enter; if left blank, an initial value determined in advance by this system will be set; the constant name described later (input with "!" Added at the beginning) can also be used; input from process computer Is also possible via a predetermined file, in which case specify "○○○ file" in the attribute field; memo Input when the user wants to leave a comment.

FIG. 12 shows an input example of the case definition editor. First, three "slabs" are defined as case facts, and "steel grade", "as the case field.
It defines four parameters, "width", "pitch", and "similarity". Data attributes are "steel grade" and "width" are integers, "pitch" and "similarity" are real numbers, and their initial values are In the flowchart showing the online execution procedure in the expert system online execution process computer 1-3 of FIG. 3, which corresponds to the case knowledge 3-1 of the past problem solving cases. You have set the definition and initial value of the set.

(2) Fact: Various knowledge phenomena used in the forward inference are expressed in the form of "~ no ~ is ~", which is input from the fact editor screen shown in FIG. The screen structure is the same as the case editor.
Fact name, field name defined as a fact,
Values, etc. are referred to in the conditional part "if ~ ~ ~ is ~" in the rule, which will be described later, and the rule execution part "~ ~ ~ ~ ~".
It is rewritten by (update, delete, generate) and used as data to control the execution of rules. FIG. 14 is an example of the input, and defines specifications such as object. The name of the fact and the initial value are set, for example, id of object is 1 and temp is 100.

(3) Constant: A fixed value such as a reference value of an event in knowledge is defined as a constant name, and it can be quoted by a constant name in other knowledge such as hypothesis facts, facts and rules. Figure 1
Input from the constant editor shown in 6. The screen structure is the same as the case definition editor. By defining constants in this editor, it is possible to maintain unity of management, and by simply changing the values on this screen, the values quoted in facts, rules, etc. can be changed all at once. Also improves. Fig. 16 shows an example of the input, "nullint" is an integer type and -999 "sup3" is a real number type, and 1200.0 is input. When quoting this value on another screen, add "!" At the beginning like "! Sup3" and quote it with a constant name.

(4) External function definition: When a user-created subroutine is desired to be called in the dynamic knowledge and dynamic case knowledge (rule) execution part, the attributes of the subroutine are shown in FIG. Defined by the function definition editor screen. The configuration of the screen is similar to that of the “case”; FIG. 18 is an example of the input, and “kadouritu” is created in C language and defines that the input has two integer arguments. "Average" is created in FORTRAN and has two arguments of real number in the input, and the result shows that the function name itself is returned in the real number type. “Locate” is created in C language and has two integer arguments and one character string argument as input, and defines that the result is returned as character string data. "Average" and "loca
te "is an external subroutine in the rule execution unit.
Be called.

Inference control knowledge input / editing (hereinafter referred to as metarule): 2-3 metarule in FIG. 2 is a dynamic knowledge or dynamic case knowledge (rule) described later, which is collected or divided into several knowledge groups. ,
This is for controlling the execution order of inference in units of the knowledge group; as a method of controlling the execution of the knowledge group, a method that is predetermined in the metarule editor screen shown in FIG. There is a method of switching the knowledge group, and both can be used together; Fig. 21 shows the meta-rule editor screen. The configuration of the screen is the same as that of the “case”, and the input items include designation of the execution order of the knowledge group (in this embodiment, execution is from top to bottom), knowledge group name, and conflict resolution method. specify. Note that the division and memo are the same as in the case editor; the knowledge group switching timing is when there is no rule that holds in the knowledge group during inference execution, or another knowledge group is called in the rule described later. FIG. 21 shows an input example thereof, and this knowledge base “test1” is composed of three knowledge groups.

Dynamic knowledge, dynamic case knowledge input editing: FIG.
2-4, the static knowledge and the case knowledge described above define the initial value of the inference, while the dynamic knowledge and the dynamic case knowledge are the knowledge that is the main body of the inference. In case-based reasoning, the form is such that the condition part "if there is a case called ~ no ~ ga ~" and the execution part "what is done about ~", and in the case of forward reasoning, the condition part "if If "of ~ is", "it is expressed in the form of" what to do with ". Also in case-based reasoning, input is made from the same rule editor screen shown in FIG. 23 (left half) and FIG. 24 (right half). 23 and 24 showing one screen are shown in FIG.
Expressed as FIG.

<Dynamic Knowledge and Dynamic Case Knowledge Input Method> The rule attributes and rule contents are defined from the rule editor screen.

(1) Rule attribute definition For each rule, the rule name, knowledge group name, priority, and
If you want to leave a comment, enter the memo field. FIG. 23 + FIG. 24 are examples of the input, the knowledge group name is “roop”, the rule name is “roop2”, and the priority is constant! lower1; (2) Rule content definition In case-based reasoning, the condition part "if there is a case called" and the execution part "what to do" are configured, and in the case of forward reasoning, the condition part "If-", the execution unit "-
What to do ". It is not possible to describe a rule without a condition part or an execution part.

<Condition part> Describe the case conditions and fact constraint conditions or search conditions necessary to execute the execution part. The execution part is executed when this condition part is satisfied. The following describes the description procedure; Fact name Describes the case facts and facts to be searched; In case-based reasoning, the description of the case function is shown; "@Case" separates the conditions that specify the case; From the second line onwards, this item may be blank if the facts are the same.

Field Name Describe case fields and fields corresponding to the above.

Index When quoting an array element by a fact, the element number is specified in the “index” column.

Value The value to be judged is described in the following format: integer, real number, character string, constant name (prefixed with "!"), Variable (prefixed with "?"), And system default value (Symbol Nulli, etc.) can be described; in the case of a variable, it means to bind (apply) a certain value to the variable; in case of case-based reasoning, the case pointer indicating the set of case facts in the first line of the case Describe variables. The case pointer variable is used to specify for which case fact set the operation described in the rule execution unit is performed. This case pointer is acquired when there is a case fact that matches the state of the case fact described in the rule condition part.

Operators "greater than or equal to", "less than or equal to", "greater than" and "less than" can be specified; it means that there is no input.

Conjunction: Specifies the connection relation between each line;
Which of “is”, “is”, “to”, “to be”, “not”, “not”, “and”, “or”, “does not exist”, and “case exists” One is specified for each line; "not", "not" means negative, and "and", "or" means connection between conditions.
"Is absent" means that the rule is satisfied if there is no fact described in the condition;
For case-based reasoning, leave the first line of the case blank,
The last line describes “There is a case”.

Calculation Formula For a variable used in the condition part and the execution part, a calculation formula for constraining the variable can be used; variable name = expression, variable name> expression, variable name <expression, variable Name> = expression, variable name <= expression, and the right-hand side is written in the usual mathematical expression using four arithmetic operations +,-, *, / and parentheses () symbols; and the right-hand side is a constant You can also use the name (add a '!' At the beginning) and the library function.

FIG. 23 is an input example of the condition part. @ Describe the conditions that specify the case in the area enclosed by the case. And the case fact set that matches this condition is a case pointer variable? Obtained by J. In this example, a case fact "slab" that is stored as a past record having a similarity to the "current slab" for which the extraction pitch is obtained is searched for as a case. As the search condition, the value within the range where the steel grade does not exceed ± 2 as defined in the calculation formula column in FIG. 24? G>? GN-2? G <? GN + 2? GN: "Current slab" steel type? G: Case fact Steel type "slab" and width is "width of current slab" + 50 or more? W>? WN + 50? WN: The width of the "current slab"? W: The width of the case fact “slab”. In the flowchart of the online execution procedure of the expert system online execution process computer 1-3 of FIG. 3, the case where search conditions are set for the current problem and a set of past problem solution cases is searched This corresponds to search 3-3.

<Execution part> Describe the contents to be executed when the condition part is satisfied. The description procedure is almost the same as the condition part, but in the operation column, “based on”, “case modification”, “case storage”, “execute”, “update”, “delete”, “create”,
The following functions can be executed by selecting and specifying "call", "display", and "start"; 1) data manipulation of facts "update" and "delete" of the value of a fact, or "generation" of a new fact. 2) Correction of facts based on case facts and storage of case facts 3) “Call” of standard subroutine prepared in the system 4) “Call” of subroutine (external subroutine) created by user 5) Fact, case fact "Display" such as data on the console 6) In addition, switch "execution" of the execution right to the knowledge group.

The description points and functions will be described below for each function; 1) Data operation of fact When the operation column is "update", the target fact name and field name (in the case of array element, specify the index) and update Specify the value. In addition, when using a variable, it is necessary to be bound by a conditional part or a calculation expression; in the case of "creation", a fact name (in the case of an array element, an index is specified) and a value are specified. When executed, new fact data is generated; in the case of “delete”, the fact name is specified; when the rule is executed, the specified fact data itself is deleted; (Deletes all related slot data); Figure 23 + Figure An example is described in 24.

2) Correction of Facts Based on Case Facts and Storing of Case Facts When the operation field is "case correction", a calculation formula field is specified in the specified case fact set (specified by the case pointer variable acquired by the condition part). The value of the fact is corrected by performing the operation specified in step 2. In the case of "case storage", the case fact is additionally stored as case knowledge based on the value calculated by the case correction; "based on" By enclosing “execute” with “execute”, a delimiter for fact modification based on case facts and storage of case facts is set; FIG. 23 (conditional part) +
An example thereof is shown in FIG. 24 (execution unit and calculation formula). In rule roop2, is the case fact set obtained in the condition part?
Specify J and select the "current slab" extraction pitch? PN is obtained by case-based reasoning and stored as new case knowledge; case-searched case fact set? J and "
The similarity? RN with the current slab "is calculated by the following formula defined in the calculation formula column;? RN = variance (? W) x 0.1 + 0.2 / deviation (? G) The first term on the right side is , The value of the variance (standard deviation) of the width? W in the case searched case fact set? J multiplied by a coefficient of 0.1; the second term on the right side is the steel type in the case searched case fact set? J. Is a value obtained by dividing the coefficient of 0.2 by the deviation (maximum value-minimum value) of? G; the case fact set? J searched for a case with the value obtained by adding the first term and the second term on the right side. And the degree of similarity with the "current slab"? RN
Calculate the extraction pitch of the "current slab"? PN is calculated by the following formula defined in the formula column :? PN = average (? P) +? RN Is this the case fact set retrieved from the cases? Pitch at J? Similarity to the average value of P? RN is added; Fig. 3
In the flowchart showing the online execution procedure in the expert computer online execution process computer cycle 1-3, the similarity between the present problem and the retrieved past problem solving case set is calculated, and the similarity is calculated by this similarity. Corresponding to the case correction 3-4 in which the setting result obtained by the case-based inference in the past problem solving case is corrected and set as the setting value of the current problem; "case storage" is described to calculate "above". The "current slab" information is set as a case fact "slab" and stored as new case knowledge. The flowchart showing the online execution procedure in the expert system online execution process computer 1-3 of FIG. 3 corresponds to the case storage 3-5 that can be used for the next case-based reasoning as a problem solving case. .

3) Call Processing of Standard Subroutine When the operation column is "Call", the following subroutines are prepared as standard subroutines in this system. For the description, specify "% standard" in the item of fact, the following subroutine name in the item of field, argument in the item of value, and "call" in the operation column.

4) Call processing of user-created external subroutine When the operation field is "call", FORTRA created by the user
It is possible to call N and C language subroutines.
For the description, specify "% external" in the item of fact, subroutine name in the item of field, argument in the item of value, and "call" in the operation column. It is necessary to define the attributes of arguments in advance in the external function editor. The argument is an integer,
Real numbers, character strings, constant names, and variables can be used. If there are multiple numbers, separate them with ',' and describe.

5) Console display of case facts, fact data, etc. When the operation field is "display", case facts and fact values can be displayed on the console window of the CRT. The description procedure specifies a case fact name and a fact name in the fact section and a case field name and a field name in the field section. Specify "display" in the operation column. The value column is unnecessary.

6) Switching the execution right to another knowledge group When the operation column is "start", the knowledge group is normally executed in the order defined by the metarule, but the knowledge group specified by the rule execution unit may be executed. I can. After the specified knowledge group is executed, the knowledge groups after the knowledge group defined by the meta-rule are sequentially executed unless changed by another rule. As a description procedure, specify the name of the knowledge group to be activated in the fact section and specify "start" in the operation field.

Although the method of inputting and editing the knowledge of the expert system using the case-based reasoning has been described above, process generation of the input and compiled knowledge, inference execution, knowledge verification, online process generation function, etc. are also available. Systemization has already been completed on a computer.

[0044]

As described above, according to the present invention, in constructing and maintaining an expert system using case-based reasoning, 1) it does not require specialized knowledge such as a language related to an advanced expert system, An expert system using case-based reasoning can be easily constructed, and its software productivity and maintainability are greatly improved. 2) By providing a case-based reasoning mechanism to conventional tools, an expert with higher application The effect is that the system can be developed. 25 and 26
An example of the development efficiency compared with a conventional system construction example using the AI language is shown in FIG.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating an expert system and a development system for controlling equipment implementing the present invention.

FIG. 2 is a flowchart showing a procedure for inputting and editing knowledge to an expert system in an expert system developing computer for carrying out the present invention.

FIG. 3 is a flowchart showing an on-line execution procedure of the case-based reasoning application expert system in the expert system online execution process computer.

[Fig. 4] CRT of a computer for expert system development
It is a top view which shows the whole display screen for the knowledge input / edit displayed on.

5 is a plan view showing a display screen when the item "editor" shown in FIG. 4 is designated. FIG.

FIG. 6 is a plan view showing a display screen when the item “utility” shown in FIG. 4 is designated.

FIG. 7 is a plan view showing a display screen when the item “knowledge base name” shown in FIG. 4 is designated.

FIG. 8 is a plan view showing a display screen when the item “knowledge base” shown in FIG. 4 is designated, showing a specific example of the display screen of the embodiment of the present invention in knowledge input / editing. Is an input example of a knowledge base editing screen.

9] The item "knowledge base copy" on the display screen of FIG.
FIG. 6 is a plan view showing a display screen when is designated.

FIG. 10: “Knowledge base registration” item on the display screen of FIG.
FIG. 6 is a plan view showing a display screen when is designated.

FIG. 11: “Delete Knowledge Base” item on the display screen of FIG.
FIG. 6 is a plan view showing a display screen when is designated.

FIG. 12 is a plan view showing a display screen when the item “case” shown in FIG. 5 is specified, and shows a specific input example of a case definition editor screen in knowledge input / editing.

FIG. 13 is a plan view showing a display screen when an attribute field shown in FIG. 12 is input.

FIG. 14 is a plan view showing a display screen when the item “fact” shown in FIG. 5 is designated, showing a specific input example of a fact editor screen in knowledge input / editing.

FIG. 15 is a plan view showing a display screen when an attribute field in FIG. 14 is to be input.

FIG. 16 is a plan view showing a display screen when the item “constant” shown in FIG. 5 is specified, and shows a specific input example of a constant editor screen in knowledge input / editing.

FIG. 17 is a plan view showing a display screen when an attribute field shown in FIG. 16 is input.

FIG. 18 is a plan view showing a display screen when the item “external function” shown in FIG. 5 is specified, and shows a specific input example of the external function editor screen in knowledge input / editing.

FIG. 19 is a plan view showing a display screen when an attribute field in FIG. 18 is to be input.

FIG. 20 is a plan view showing a display screen when an attribute field in FIG. 18 is to be input.

FIG. 21 shows a screen displayed when the item “meta rule” shown in FIG. 5 is specified, and shows a specific input example of a constant editor screen in knowledge input / editing.

22 is a plan view showing a display screen when an attempt is made to input a conflict resolution field in FIG. 21. FIG.

FIG. 23 is a plan view showing the left half of the screen displayed by designating the item “rule” shown in FIG. 5 and further designating it from the rule list displayed by the rule list. Is a concrete example of input of.

FIG. 24 is a plan view showing the right half of the screen displayed by designating the item “rule” shown in FIG. 5 and further designating from the rule list displayed thereby, and is the rule editor screen in knowledge input / editing. Is a concrete example of input of. The whole one screen is completed by superposing the dividing line (virtual line) DL of FIG. 23 and the dividing line DL of FIG.

FIG. 25 is a graph showing the development period of the conventional system construction in the AI language and the system construction according to the present invention, showing the case where the “development tool” in the figure uses the present invention.

FIG. 26 is a graph showing the development period of the conventional system construction in the AI language and the system construction according to the present invention, showing the case where the “development tool” in the figure uses the present invention.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Isao Otani 20-1 Shintomi, Futtsu City, Chiba Shin Nippon Steel Co., Ltd. Technology Development Division

Claims (1)

[Claims] 1. An attempt is made to solve a knowledge representation method of an expert system using case-based reasoning, inference control knowledge, dynamic knowledge, static knowledge, case knowledge which is a set of past problem solving cases, and present knowledge. A general-purpose system for inputting knowledge about 5 types of dynamic case knowledge, which retrieves case knowledge with high similarity to the problem, corrects it according to the degree of similarity, and stores the result as a new problem solving case. A knowledge input method for case-based reasoning, in which a tabular editor screen is displayed, and the information entered in the fields displayed on the editor screen is read and edited.