JP2010250698A - Production rule-generating device and program thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To generate a production rule having a conditions part which designates agreeing conditions, disagreeing conditions and numerical-value ranges, on the basis of input information on conditional values which correspond to conditions items and which do not cause troublesomeness in functional alteration, along with the addition or alteration of items to the conditions part or a conclusion part. <P>SOLUTION: Information for conditions part-setting which includes the combination of an rule ID and conditional values, which are correlated with a plurality of conditions items, respectively, is classified into four kinds of information; that is, information for agreeing conditions-setting I1, information for disagreeing conditions-setting I2, information for lower-limit conditions-setting I3 and information for upper-limit conditions-setting I4, and is then inputted. The rule ID sets, for the same information, conditions parts which regard agreement or disagreement of the conditions items with the conditional values of the information I1 as the conformity conditions, on the basis of the contents of the set conditional values of the information I1 and information I2 (S13), and also sets conditions parts which regard the numerical value ranges as conformity conditions, on the basis of the contents of the set conditional values of the information I3 and information I4 (S14). <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a production rule generation device that generates a production rule used in a rule-based system based on input information, and a program therefor.

The rule-based system is an effective means for utilizing various business knowledge using a computer. The rule-based system includes means for storing a set of production rules including a condition part and a conclusion part, and an inference engine that performs a process for deriving a conclusion based on the production rule and separately provided condition information. The inference engine determines whether the condition part and the condition information match in the production rule. If the condition part and the condition information match, the inference engine infers the content of the conclusion part in the production rule. Output as a conclusion. In general, production rules in a rule-based system are added or modified as needed.
Patent Document 1 and Patent Document 2 show an example of a production rule semantic interpretation method in a rule-based system. Patent Document 3 discloses an example of a method for adjusting a numerical range of a condition part in a production rule.

By the way, in a rule-based system for uses such as design support, items of a condition part and a conclusion part in a production rule are often selected from a plurality of predetermined candidates. Hereinafter, a production rule of a type in which items in the condition part and the conclusion part are selected from a plurality of predetermined candidates is referred to as an item selection type production rule.
In the item selection type production rule, for example, all possible customer requirement specification items are candidates for the condition part, and all possible manufacturing condition items are candidates for the conclusion part. Hereinafter, the item of the condition part in the production rule is referred to as a condition item, and the item of the conclusion part is referred to as a conclusion item.
Therefore, the apparatus for setting the item selection type production rule can basically set the production rule only by obtaining the following rule setting information. That is, the rule setting information is information on a condition value and a conclusion value associated with one or more of a plurality of items for each of the condition item and the conclusion item. Such rule setting information includes, for example, inputting the condition value or the conclusion value in a spreadsheet whose title row is a plurality of cells in which a plurality of items of the condition item and the conclusion item are preset. Can be easily obtained. Based on the rule setting information thus obtained, a production rule that “if the condition item matches the condition value, the conclusion value is adopted for the conclusion item” can be easily set. Hereinafter, such a production rule is referred to as a matching condition designation type production rule.

In addition to the matching condition designation type, the following other types of item selection type production rules may be considered.
For example, a production rule may be one in which the negative condition part is set, such as “If the condition item does not match the condition value, the conclusion value is adopted for the conclusion item”. Hereinafter, this is referred to as a production rule of a mismatch condition designation type. In addition, the production rule defines a numerical range as the condition condition, such as “If the value of the condition item is within a range from a predetermined lower limit value to a predetermined upper limit value, the conclusion value is adopted for the conclusion item”. What is included in the part is also conceivable. Hereinafter, this is referred to as a numerical range designation type production rule.

JP-A-2-193225 JP 2008-234011 A JP 2001-312408 A

By the way, the information necessary for setting the non-matching condition designating type production rule and the numerical range designating type production rule is not sufficient if only the condition item and the conclusion item are associated with the condition value and the conclusion value. Absent. That is, the conventional apparatus for setting the non-matching condition specifying type production rule and the numerical value range specifying type production rule has a special type for specifying the negative condition part and the condition part including the numerical value range. It was necessary to have a man-machine interface.
However, in a rule-based system for applications such as design support, the number of candidates for the condition item and the conclusion item may reach several hundred to several thousand, and the addition or change of the item candidate also occurs at any time. The apparatus for setting the item selection type production rule specifies the condition part including the negative condition part and the numerical value range according to addition or change of a large number of candidates for the condition item and the conclusion item. It is necessary to update the man-machine interface for this purpose. There was a problem that such an update was very troublesome.
Accordingly, the present invention has been made in view of the above circumstances, and the object of the present invention is to provide a matching condition designation type based on input information of values associated with items of a condition part and a conclusion part in a production rule. Therefore, it is an object of the present invention to provide a production rule generation apparatus and program for generating a production rule of a mismatch condition specification type and a numerical value range specification type, and which does not require the trouble of changing functions due to the addition or change of the items.

In order to achieve the above object, a production rule generation device according to the present invention is a device that generates a production rule used in a rule-based system based on predetermined input information, and includes the following (1) to (3) and ( Each component shown in (3-1) to (3-4) is provided.
(1) Condition part setting information comprising a combination of rule identification information for uniquely identifying each of the production rules and condition values associated with a plurality of preset condition items for the condition part of the production rule At least the first condition part setting information for setting the matching condition, the second condition part setting information for setting the mismatch condition, the third condition part setting information for setting the numerical range lower limit condition, and the numerical range upper limit condition setting. The condition part setting information input means for distinguishing and inputting into four types of the fourth condition part setting information.
(2) Conclusion part setting information for inputting conclusion part setting information composed of a combination of the rule identification information and a conclusion value associated with each of a plurality of preset conclusion items for the conclusion part of the production rule Input means.
(3) For each combination of the four types of the condition part setting information and the conclusion part setting information having the same rule identification information, the condition part and the conclusion part are set based on the combination. Production rule automatic generation means for automatically generating production rules and storing them in the storage means.
Here, the production rule automatic generation means includes the components shown in the following (3-1) to (3-4).
(3-1) In the first condition part setting information and the second condition part setting information having the same rule identification information, the condition value is present in the first condition part setting information, and The condition item for which the condition value does not exist in the second condition part setting information is determined, and the condition is that the condition item matches the content corresponding to the condition value of the first condition part setting information. Matching condition part setting means for setting the condition part of the production rule.
(3-2) The condition item in which the condition value exists in both the first condition part setting information and the second condition part setting information having the same rule identification information is determined, and the condition item A mismatch condition part setting means for setting the condition part of the production rule whose conforming condition is that it does not match the content corresponding to the condition value of the first condition part setting information.
(3-3) With respect to the condition item in which the condition value exists in at least one of the third condition part setting information and the fourth condition part setting information having the same rule identification information, the third condition part Numerical value range condition part setting means for setting the condition part of the production rule with a numerical value range having a numerical value corresponding to the condition value in the setting information and the fourth condition part setting information as a threshold value.
(3-4) The conclusion item in which the condition value exists in the conclusion part setting information having the same rule identification information is determined, and the conclusion item corresponds to the conclusion value of the conclusion part setting information. Conclusion part setting means for setting the conclusion part of the production rule whose content is a conclusion.

In the present invention, the four types of condition part setting information and conclusion part setting information are information on values (the condition value or the conclusion value) associated with the condition part and the conclusion part items in the production rule. It is information to which the rule identification information is added. These four types of condition part setting information and conclusion part setting information include, for example, a title row of a cell column in which a title of the rule identification information and a plurality of the condition items or the conclusion items are preset. Can be easily obtained by inputting the rule identification information, the condition value, or the conclusion value.
The four types of condition part setting information and conclusion part setting information are associated with each content to be reflected in one rule by the rule identification information.
The four types of condition part setting information are information for designating the following four types of contents.
That is, the first condition part setting information is information for designating the value of each condition item in the condition part of the production rule of the matching condition designation type or the mismatch condition designation type.
The second condition part setting information is information for designating whether the condition part of the production rule is the matching condition designation type or the mismatch condition designation type.
Further, the third condition part setting information and the fourth condition part setting information are information for specifying the lower limit value and the upper limit value of the numerical value range in the numerical value range specification type production rule, respectively.
The production rule generation device according to the present invention inputs the information for setting the condition part into four types in order to be able to specify the type and contents of the condition part of the production rule. Here, when the information for setting the condition part is classified into four types, it becomes a problem how to classify the information for each production rule that should reflect them. In the present invention, the problem is solved by the rule identification information. In other words, the production rule generation device according to the present invention classifies the information for setting the condition part and the information for setting the conclusion part, which are classified into four types, into pieces that match the rule identification information. To reflect.
As described above, according to the production rule generation device of the present invention, based on the input information of values associated with the condition part and conclusion part items in the production rule, the matching condition designation type, the mismatch condition designation type, and the numerical range A production rule of a specified type can be generated. In addition, there is no need to change functions due to the addition or change of the items.

In the present invention, as the condition part setting information input means and the conclusion part setting information input means, a first example having the configuration shown in the following (1-1) and (2-1), , (1-2) and a second example having the configuration shown in (2-2), a third example having a configuration shown in (1-3) and (2-3), (1-4) And the 4th example provided with the composition shown by (2-4) etc. can be considered.
The first example is shown below.
(1-1) For each of the first condition part setting information to the fourth condition part setting information, the condition part setting information input means displays information on a plurality of the condition items on a predetermined display means. Individually, the condition value corresponding to each of the condition items and the rule identification information are input in association with each other according to an input operation through a predetermined operation unit.
(2-1) The conclusion part setting information input means displays information on a plurality of conclusion items on the display means, and corresponds to each conclusion item according to an input operation through the operation part. The conclusion value and the rule identification information are input in association with each other.
The second example is shown below.
(1-2) The condition part setting information input unit responds to each of the condition items in accordance with an input operation through a predetermined operation unit while displaying information on a plurality of the condition items on a predetermined display unit. The condition value, the rule identification information, and the condition identification information for identifying which of the first condition part setting information to the fourth condition part setting information corresponds to each of the condition value and the rule identification information.
(2-2) The conclusion part setting information input means displays information on a plurality of conclusion items on the display means, and corresponds to each conclusion item according to an input operation through the operation part. The conclusion value and the rule identification information are input in association with each other.

The third example is shown below.
(1-3) The condition part setting information input unit reads the information from the predetermined storage unit or receives the information from the external device through the predetermined communication unit, thereby receiving the first condition unit setting information to the first unit. Each of the four condition part setting information is individually input.
(2-3) The conclusion part setting information input unit inputs the conclusion part setting information by reading information from a predetermined storage unit or receiving information from an external device through a predetermined communication unit.
The fourth example is shown below.
(1-4) The condition part setting information input means reads out information from a predetermined storage means or receives information from an external device through a predetermined communication means, and selects one of the information of the plurality of condition items. The combination information of the condition value, the rule identification information and the condition distinguishing information for distinguishing between the first condition part setting information and the fourth condition part setting information is input. .
(2-4) The conclusion part setting information input unit inputs the conclusion part setting information by reading information from a predetermined storage unit or receiving information from an external device through a predetermined communication unit.
The present invention can also be understood as a production rule generation program for causing a computer to execute the processing of each means in the production rule generation apparatus according to the present invention described above.

  According to the present invention, a production rule of a matching condition designation type, a mismatch condition designation type, and a numerical range designation type is generated based on input information of values associated with items of a condition part and a conclusion part in the production rule. Therefore, it is possible to provide a production rule generation device and a program thereof that do not require the trouble of changing the function due to the addition or change of the item.

The schematic block diagram of the design rule registration system A containing the design rule production | generation apparatus X which is an example of the production rule production | generation apparatus which concerns on embodiment of this invention. 1 is a schematic configuration diagram of a design rule generation device X. FIG. 5 is a flowchart showing an example of a procedure of design rule registration processing by the design rule generation device X. 6 is a flowchart showing an example of a procedure of design rule automatic generation processing by the design rule generation device X. The figure showing the 1st example of the information input screen for rule settings which the design rule production | generation apparatus X displays on a terminal. The figure showing an example of the content of the design rule produced | generated by the design rule production | generation apparatus X. The figure showing the 2nd example of the information input screen for rule settings which the design rule production | generation apparatus X displays on a terminal.

  Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings so that the present invention can be understood. The following embodiment is an example embodying the present invention, and does not limit the technical scope of the present invention.

First, a schematic configuration of a design rule generation apparatus X which is an example of a production rule generation apparatus according to an embodiment of the present invention and a design rule registration system A including the same will be described with reference to FIGS. 1 and 2.
As shown in FIG. 1, the design rule registration system A includes the design rule generation device X and a terminal 2, which are configured to communicate with each other through a network 1 such as an intranet or the Internet.
The design rule generating device X receives information for setting production rules from the terminal 2 via the network 1, and generates a design rule for design support used in a rule-based system based on the input information. It is.
On the other hand, the terminal 2 displays information input screens on the display according to the information on the production rule setting information input screen obtained from the design rule generation device X via the network 1, and displays information such as a keyboard and a mouse. Information for setting production rules is input through the input means. Further, the terminal 2 transmits the input production rule setting information to the design rule generation device X.

As shown in FIG. 2, the design rule generation device X includes a calculation unit 11, an operation display unit 12, a communication control unit 13, a storage unit 14, and the like.
The calculation unit 11 is a calculation unit including a CPU, a ROM, a RAM, and the like, and executes a design rule registration program and other programs described later.
The operation display unit 12 includes a display device such as a liquid crystal display (not shown) and an input device such as a keyboard and a mouse, and is a man-machine interface that performs various information input processing and display processing in the design rule generation device X. .
The communication control unit 13 is a communication interface that performs communication control based on a predetermined communication protocol (TCP / IP or the like) with the terminal 2 via the network 1.
The storage unit 14 is a storage device such as a hard disk, and stores various programs executed by the arithmetic unit 11 and various data referred to or recorded when the programs are executed.
In the present embodiment, a design rule registration program p1 and a web server program p2 are stored in advance in the storage unit 14 as programs executed by the calculation unit 11. Further, rule setting information d1 that is input and referred to when the design rule registration program p1 is executed, and design rule information d2 that is automatically generated are also stored in the storage unit 14.

On the other hand, the terminal 2 is a computer such as a general personal computer. Specifically, the terminal 2 includes, as components not shown, a man-machine interface including a display device and an input device, an arithmetic unit including an MPU, a main memory, and peripheral devices, a hard disk, and a nonvolatile memory. A data storage unit such as a random access memory, and a communication control unit that performs data communication via the network 1.
The terminal 2 is installed with a web browser program. The process of inputting the production rule setting information to be transmitted to the design rule generating device X while displaying the information input screen on the display device is executed through a web browser.
Specifically, the calculation unit 11 in the design rule generation device X generates screen data for inputting information for setting production rules by executing the web server program p2 to the terminal 2. Send. The calculation unit 11 receives information returned in response to the transmission of the screen data. Each screen data is data described in, for example, HTML or JavaScript (registered trademark of Sun Microsystems, Inc., the same applies hereinafter).

Next, the procedure of the design rule registration process by the design rule generation device X will be described with reference to the flowcharts shown in FIGS.
In the design rule generation device X, the calculation unit 11 performs the following processing by executing the design rule registration program p1 and the web server program p2. The calculation unit 11 executes the following process when a request for starting a design rule registration process is received from the terminal 2. S1, S2,... Shown below represent identification codes of processing steps (steps) performed by the arithmetic unit 11 by executing the design rule registration program p1.
First, the calculation unit 11 executes a process for causing the terminal 2 to output a predetermined rule setting information input screen (S1). Specifically, the calculation unit 11 transmits the input screen data of the rule setting information to the terminal 2 that is the transmission source of the start request by executing the web server program p2.
Further, the calculation unit 11 receives the rule setting information input from the terminal 2 (S2). Hereinafter, the terminal 2 that is a communication partner of the design rule generation device X is a transmission source of the start request.

FIG. 5 is a first example of a rule setting information input screen displayed on the terminal 2. This input screen is a screen displayed on the display device by the web browser of the terminal 2 based on the input screen data transmitted from the design rule generating device X to the terminal 2 in step S1.
As shown in FIG. 5, the rule setting information input screen specifies the rule condition part setting information d11 that is information for specifying the contents of the production rule condition part and the contents of the conclusion part of the production rule. It is a screen for inputting rule conclusion part setting information d12, which is information for performing, in a table format.
The rule condition part setting information d11 includes matching condition setting information I1, negative condition setting information I2 for setting a mismatch condition, lower limit condition setting information I3 defining a lower limit condition of a numerical range, and a numerical value. The upper limit condition setting information I4 that defines the upper limit condition of the range is classified into four types.
Each of the four types of information I1 to I4 in the rule condition part setting information d11 is associated with a rule ID (I0) for uniquely identifying each production rule and each of a plurality of preset condition items h1. The condition values I11, I12, I21, I31, and I41 are input as combinations. That is, the rule condition part setting information d11 includes a plurality of sets of information in which one rule ID (I0) and one or more condition values I11, I12, I21, I31, and I41 are associated with each other. Input as set information.

There are two types of condition values I11 and I12 that can be set in the matching condition setting information I1, a direct specification value I11 that directly specifies a value and an indirect specification value I12 that includes a wild card. For example, the wild card “%” in the indirectly specified value I12 shown in FIG. 5 means NULL or an arbitrary character string of one or more characters. Therefore, “% C” shown in FIG. 5 represents an arbitrary character string such as “C”, “AC”, “ABC”, “ABDC”, or the like that ends with “C”. In addition to “%”, “?” Representing an arbitrary character can be set as a wild card.
Further, the condition value I21 that can be set in the negative condition setting information I2 is a character "!" Indicating that it is a negative condition. If "!" Is not set as the condition value I21 of the negative condition setting information I2, it means that the condition is positive.
The condition value I31 that can be set in the lower limit condition setting information I3 and the condition value I41 that can be set in the condition condition setting information I4 are numerical values. Needless to say, when both the condition value I31 and the condition value I41 are set, the condition value I41 is set to be larger than the condition value I31.

On the other hand, the rule conclusion part setting information d12 is similar to the rule condition part setting information d11 in that the rule ID (I0) and the conclusion value I5 associated with each of the plurality of preset conclusion items h2 are set. As a combination. That is, the rule conclusion part setting information d12 is input as information including one set of information in which one rule ID (I0) and one or more conclusion values I5 are associated with each other. The
As shown in FIG. 5, in the rule setting information input screen, a plurality of cells in which a title “rule ID” of the rule ID (I0) and a plurality of the condition items h1 are set in advance are used as title rows. This is a screen for displaying a spreadsheet (table), and each cell in the spreadsheet is an input frame for the rule ID (I0) or the condition values I11, I21, I31, I41.
Further, the rule setting information input screen displays a spreadsheet (table) having a plurality of cells in which the title “rule ID” of the rule ID (I0) and the plurality of conclusion items h2 are set in advance as title rows. It is also a screen to be displayed, and each cell in the spreadsheet is an input frame for the rule ID (I0) or the conclusion value I5.

As shown in FIG. 5, the four types of condition part setting information I1 to I4 and the rule conclusion part setting information d12 correspond to each content to be reflected in one rule by the rule ID (I0). Attached.
The matching condition setting information I1 is information for designating the value I11 or I12 of each condition item in the condition part of the production rule of the matching condition designation type or the mismatch condition designation type.
The negative condition setting information I2 is information for designating whether the condition part of the production rule is the matching condition designation type or the mismatch condition designation type.
The lower limit condition setting information I3 and the upper limit condition setting information I4 are information for specifying the lower limit value and the upper limit value of the numerical value range in the numerical value range specification type production rule, respectively.
The terminal 2 functions as input means for the rule condition part setting information d11 and the rule conclusion part setting information d12 for the design rule generation device X.
In the example shown in FIG. 5, the terminal 2 displays the information of the plurality of condition items h1 on a predetermined display device, and individually displays the four types of condition part setting information I1 to I4. The condition values I11, I12, I3, I4 and the rule ID (I0) corresponding to each of the condition items h1 are input in association with each other according to an input operation through an operation unit such as a keyboard or a keyboard (condition setting) Example of information input means).
Furthermore, the terminal 2 displays the information of the plurality of conclusion items h2 on the display device, and according to the input operation through the operation unit, the conclusion value I5 corresponding to each of the conclusion items h2 and the rule ID (I0) is input in association with it.

The rule setting information input screen includes rule generation buttons B1, which are three types of operation buttons for generating a trigger for transmitting the input information and processing instructions to the design rule generation device X. A save button B2 and a cancel button B3 are included.
When the rule generation button B1 is operated, information input through the rule setting information input screen and a command for requesting automatic generation of a design rule based on the information are sent from the terminal 2 to the design rule generation device. Sent to X.
When the save button B2 is operated, the information input through the rule setting information input screen and a command for requesting that the information be stored in the storage unit 14 in the design rule generation device X are: The data is transmitted from the terminal 2 to the design rule generation device X.
When the cancel button B3 is operated, a command for requesting to stop the process is transmitted from the terminal 2 to the design rule generation device X.
Hereinafter, information combining the rule condition part setting information d11 and the rule conclusion part setting information d12 is referred to as rule setting information d1.

And the said calculating part 11 in the said design rule production | generation apparatus X will receive the instruction | indication according to the said rule setting information and the said operation button B1-B3 input in the said terminal 2 (S2), The content of the instruction | command will be shown. Discriminate (S3, S5).
When the command is a cancel request command corresponding to the cancel button B3, the calculation unit 11 ends the processing as it is.
On the other hand, when the command is a rule automatic generation request corresponding to the rule generation button B1 or a storage request command corresponding to the save button B2, the calculation unit 11 receives the rule setting information received from the terminal 2. d1 is stored in the storage unit 14 (S4).
Further, when the command is the command for automatic generation request, the calculation unit 11 executes a design rule automatic generation process S6 based on the rule setting information d1 received from the terminal 2 (S6). Thereafter, the calculation unit 11 ends the design rule registration process.
On the other hand, when the command is the command for the storage request, the calculation unit 11 ends the design rule registration process after the process of step S4.

Next, details of the design rule automatic generation processing executed in step S6 will be described with reference to the flowchart of FIG.
First, the calculation unit 11 sets a variable i representing the rule ID (I0) to an initial value (i = 1) (S10). Thereafter, the calculation unit 11 sequentially increments the variable i (S19), and for each set of information having the same rule ID (I0) in the rule setting information d1, the following steps S11 to S11 are performed. The process of S18 is repeated.
That is, the arithmetic unit 11 sets a variable j representing the number of the condition item h1 to an initial value (j = 1) (S11).
Next, the calculation unit 11 has the rule ID (I0) set to i from the rule setting information d1 obtained in step S2, and is associated with the jth condition item. A set of the condition values is extracted (S12). Hereinafter, the information extracted in step S12 is referred to as extraction condition information Dx (i, j).
For example, when the content of the rule condition part setting information d11 is the content shown in FIG. 5 and i = 1 and j = 3, the extraction condition information Dx (i, j) Each condition value for the condition item h3 = “dimension” in the information in which the rule ID (I0) is set to “0001”, that is, “3” and “2” that are the condition value I1 of the matching condition is negated. The condition value I2 of the condition “!” Is included.
When i = 2 and j = 3, the extraction condition information Dx (i, j) includes the condition item h1 = ”dimension in the information in which the rule ID (I0) is set to“ 0002 ”. "20" that is the condition value I3 of the lower limit condition and "40" that is the condition value I4 of the upper limit condition are included.
In the extraction condition information Dx (i, j), each of the condition values includes the matching condition setting information I1, the negative condition setting information I2, the lower limit condition setting information I3, and the upper limit condition setting information. A distinction is made between I4 and I4.

Next, the calculation unit 11 performs the following procedure based on the condition values I11 and I21 of the matching condition setting information I1 and the negative condition setting information I2 in the extraction condition information Dx (i, j). , A matching condition or a mismatching condition is set for the j-th condition item h1 in the condition part of the i-th design rule (S13).
That is, the calculation unit 11 determines whether or not the content of the extraction condition information Dx (i, j) corresponds to each of the following first setting state and second setting state.
In the first setting state, the condition value I11 or I12 of the matching condition setting information I1 exists and the condition value I21 of the negative condition setting information I2 exists for the jth condition item h1. It is a state that does not.
The second setting state includes both the condition value I11 or I12 of the matching condition setting information I1 and the condition value I21 of the negative condition setting information I2 for the jth condition item h1. It is a state to do.
When the calculation unit 11 determines that the content of the extraction condition information Dx (i, j) is in the first setting state, the calculation unit 11 sets the matching condition for the jth condition item h1. The condition part of the design rule (production rule) having a matching condition corresponding to the content corresponding to the condition value I11 or I12 of the information I1 is set (S13).
On the other hand, when it is determined that the content of the extraction condition information Dx (i, j) is in the second setting state, the calculation unit 11 sets the matching condition for the jth condition item h1. The condition part of the design rule having a matching condition that does not match the content corresponding to the condition value I11 or I12 of the information I1 is set (S13).
Here, when the condition values I11 and I12 are set in a plurality of the condition items h1 for the same rule ID (I0), they are set as AND conditions. On the other hand, when a plurality of the condition values I11 and I12 are set in one condition item h1 for the same rule ID (I0), the condition value I2 (“!”) Of the negative condition corresponding thereto is set. Unless are set, they are set as OR conditions. However, for the same rule ID (I0), a plurality of the condition values I11 and I12 are set in one condition item h1, and the corresponding condition value I2 (“!”) Of the negative condition is set. In this case, the plurality of condition values I11 and I12 are set as AND conditions.

For example, when the content of the rule condition part setting information d11 is the content shown in FIG. 5, and i = 1 and j = 1, the condition value “!” Is included in the negative condition setting information I2. Since it does not exist, the condition part of the first design rule is set with a conforming condition that “standard” matches “AA”, or “standard” matches “BB”, that is, a matching condition. Is done.
Furthermore, when i = 1 and j = 3, since the condition value “!” Exists in the negative condition setting information I2, the “dimension” is “3” in the condition part of the first design rule. A conformance condition that does not match “and“ dimension ”does not match“ 2 ”, that is, a mismatch condition is set. These two matching conditions are set as AND conditions.

Next, the calculation unit 11 determines the i-th design rule based on the condition values I31 and I41 of the lower limit condition setting information I3 and the upper limit condition setting information I4 in the extraction condition information Dx (i, j). A numerical range condition is set for the j-th condition item h1 in the condition part (S14).
For example, when the content of the rule condition part setting information d11 is the content shown in FIG. 5 and i = 2 and j = 3, the lower limit condition setting information I3 and the upper limit condition setting information Since “20” and “40” are set as the respective condition values I31 and I41 of I4, the “dimension” is “20” or more and “dimension” in the condition part of the second design rule. Is set to a matching condition in a numerical range of “40” or less.
Then, the arithmetic unit 11 counts up the value of the variable j (S16), and performs the above-described condition part setting processing (S12 to S14) for each of the condition items h1 in one design rule. The calculation unit 11 counts up the value of the variable j (S16) until it is determined that the setting processing (S12 to S14) of the condition units related to all the condition items h1 is completed for each design rule (S15). ), Condition part setting processing (S12 to S14).

In the rule condition part setting information d11, for the combination of the same rule ID (I0) and the same condition item h1, the condition value I31 of the lower limit condition setting information I3 and the upper limit condition setting information I4 A case where only one of the condition values I41 is set is conceivable. In this case, when only the condition value I31 of the lower limit condition setting information I3 is set, the calculation unit 11 sets the corresponding condition item h1 to be equal to or greater than the condition value I31. The condition part is set. Similarly, when only the condition value I41 of the upper limit condition setting information I4 is set, the calculation unit 11 sets the corresponding condition item h1 to be equal to or less than the condition value I41. The condition part is set.
That is, the calculation unit 11 includes the condition item h1 in which the condition values I31 and I41 exist in at least one of the lower limit condition setting information I3 and the upper limit condition setting information I4 having the same rule ID (I0). With respect to the above, the condition part is set with the numerical range having threshold values corresponding to the condition values I31 and I41 in the lower limit condition setting information I3 and the upper limit condition setting information I4.

In addition, when the setting process (S12 to S14) of the condition part for all the condition items h1 is completed for each design rule, the calculation part 11 executes the setting process of the conclusion part in the design rule (S17).
Further, the calculation unit 11 stores the design rule information d2 that is a set of design rules in which the condition part and the conclusion part are set in the storage unit 14 (S18).
In step S17, first, the calculation unit 11 obtains a set of conclusion values for each conclusion item in which the rule ID (I0) is set to i from the rule setting information d1 obtained in step S2. Extract.
Next, the calculation unit 11 concludes that the content of the conclusion item is the conclusion value corresponding to the conclusion item for the conclusion item for which the conclusion value is set based on the extracted information. Set the part.
For example, when the content of the rule conclusion part setting information I2 is the content shown in FIG. 5 and i = 1, “2” for each of “heat treatment time” and “heat treatment temperature” in the conclusion item h2. Since the conclusion values of “and“ 650 ”are set, the conclusion part of the first design rule includes“ “heat treatment time” as “2” hours and “heat treatment temperature” as “650 ° C.”. The conclusion part is set. Similarly, when i = 2, the conclusion values “1” and “600” are set for the “heat treatment time” and “heat treatment temperature” in the conclusion item h2, respectively. In the conclusion part, a conclusion part is set that “the heat treatment time” is “1” time and the “heat treatment temperature” is “600 ° C.”.

The calculation unit 11 counts up the value of the variable i (S19), and for each design rule, the condition part setting process (S11 to S16), the conclusion part setting process (S17), and the design A rule information storage process (S18) is performed. The calculation unit 11 counts up the value of the variable i (S19) until it determines that the setting process (S11 to S19) of the condition part and the conclusion part regarding all design rules has been completed (S20). And the setting process (S11-S17) of a conclusion part and the preservation | save process (S18) of the information of the design rule obtained by it are performed. The number of all design rules is determined by the number of types of the rule ID (I0) set in the rule setting information d1.
FIG. 6 is a diagram showing the contents of the design rule automatically generated by the design rule generation device X based on the rule setting information d1 shown in FIG.
The design rule information d2, which is a set of production rules generated and stored by the design rule generation device X, is referred to by an inference device in a predetermined rule base system and used as a rule base for inference processing. That is, the inference apparatus determines whether or not a predetermined input value corresponding to the condition item h1 matches the condition part of each design rule in the design rule information d2, and when determining that it matches, A combination of the conclusion item and the conclusion value set in the conclusion part of the rule is derived as an inference result. Since inference processing based on production rules is well known, detailed description thereof is omitted here.

As described above, the calculation unit 11 in the design rule generation device X uses the four types of condition part setting information I1 to I4 and the rule conclusion part setting information having the same rule ID (I0). For each combination of information d12, the design rule that is one production rule in which the condition part and the conclusion part are set based on the combination is automatically generated (S11 to S17), and the information on the design rule is stored in the memory The design rule is automatically stored in the unit 14 (S18).
Further, the calculation unit 11 adds the condition value I11 or I12 to the matching condition setting information I1 in the matching condition setting information I1 and the negative condition setting information I2 having the same rule ID (I0). And the condition item h1 in which the condition value I21 does not exist in the negative condition setting information I2 is determined (S13). Further, the calculation unit 11 sets a condition part of the design rule having a matching condition that the condition item h1 matches the content corresponding to the condition value I11 or I12 of the matching condition setting information I1 ( S13: Matching condition part setting process).
In addition, the calculation unit 11 includes the condition item h1 in which the condition value I11 or I12 exists in both the matching condition setting information I1 and the negative condition setting information I2 having the same rule ID (I0). Is discriminated (S13). Further, the calculation unit 11 sets a condition part of the design rule with a condition that the condition item h1 does not match the content corresponding to the condition value I11 or I12 of the matching condition setting information (S13). : Mismatch condition part setting process).
In addition, the calculation unit 11 includes the condition item h1 in which the condition values I31 and I41 exist in both the lower limit condition setting information I3 and the upper limit condition setting information I4 having the same rule ID (I0). Is discriminated (S14). Further, the calculation unit 11 sets a numerical value corresponding to the condition value I31 of the lower limit condition setting information I3 and a numerical value corresponding to the condition value I41 of the upper limit condition setting information I4 for the condition item h1. A condition part of the design rule having a condition that it is within a numerical value range as an upper limit value is set (S14: numerical value range condition part setting process).
In addition, the calculation unit 11 determines the conclusion item h2 in which the condition value I5 exists in the rule conclusion part setting information d12 having the same rule ID (I0) (S17). Further, the computing unit 11 sets the conclusion part of the design rule that concludes the content corresponding to the conclusion value I5 of the rule conclusion part setting information d12 with respect to the conclusion item h2 (S17: conclusion part setting). processing).

In the design rule generation device X, the four types of condition part setting information I1 to I4 and the rule conclusion part setting information d12 are values associated with the condition part and the conclusion part items h1 and h2 in the production rule. This is information in which the rule ID (I0) is added to the information of I11, I12, I21, I31, I41, and I5. The four types of condition part setting information I1 to I4 and the rule conclusion part setting information d12 include, for example, a title of the rule ID (I0) and a plurality of the condition items as shown in FIG. The rule ID (I0), the condition values I11, I12, I21, I31, I41 or the conclusion value I5 are input to a spreadsheet whose title row is a cell column in which h1 or the conclusion item h2 is preset. Can be easily obtained.
Further, in the design rule generation device X, the rule condition part setting information d11 is divided into four types of information I1 to I4 and input in order to be able to specify the type and contents of the condition part of the production rule. To do. Here, when the rule condition part setting information d11 is classified into four types, it becomes a problem how to classify them for each production rule that should reflect them. In the design rule generation device X, the problem is solved by the rule ID (I0). That is, the design rule generation device X classifies the condition part setting information I1 to I4 and the conclusion part setting information d12 divided into four types according to the rule IDs (I0) that match. Reflect in one production rule.
As described above, according to the design rule generation device X, based on input information of values associated with the condition part and the conclusion part items in the production rule, the matching condition designation type, the mismatch condition designation type, and the numerical range designation type Production rules can be generated. In addition, there is no need to change functions due to the addition or change of the items.

Next, a second example of an input screen for the rule condition part setting information d11 and the rule conclusion part setting information d12 through the terminal 2 will be described with reference to FIG.
In the example shown in FIG. 7, the terminal 2 displays the information of the plurality of condition items h1 on a predetermined display device, and the condition values I11, I12, I3, I4 corresponding to each of the condition items h1 and , The rule ID (I0) and the condition distinction ID (I6) for distinguishing which of the four types of condition part setting information I1 to I4 corresponds to each other (input of condition part setting information) Example of means).
Furthermore, the terminal 2 displays the information of the plurality of conclusion items h2 on the display device, and according to the input operation through the operation unit, the conclusion value I5 corresponding to each of the conclusion items h2 and the rule ID (I0) is input in association with it. Note that the values “1”, “2”, “3”, and “4” of the condition identification ID (I6) shown in FIG. 7 are the matching condition setting information I1, the negative condition setting information I2, and the like, respectively. This corresponds to the lower limit condition setting information I3 and the upper limit condition setting information I4.
Then, the design rule generation device X, in steps S12 to S14 and S17 described above, based on the condition identification ID (I6), the matching condition setting information I1, the negative condition setting information I2, and the lower limit condition. The setting information I3 and the upper limit condition setting information I4 are distinguished.
A process of inputting the rule condition part setting information d11 and the rule conclusion part setting information d12 through the input screen as shown in FIG.

  In addition to the case where the design rule generation device X inputs the rule condition part setting information d11 by distinguishing it from the above-described four types of information I1 to I4, the design rule generation device X further displays other types of information exemplified below. It is also conceivable to input the information separately including five or more types of information. As the information after the fifth type, for example, for the condition item h1 in which the condition values I11, I12, I21, I31, and I41 are set, information having priority as the condition value when the condition is satisfied or priority Information with a degree as a condition value can be considered. In this case, it is conceivable that the inference device controls the output order of the inference results based on the fitness level and the priority level.

  The present invention can be used for an apparatus that generates a production rule used in a rule-based system based on input information.

A: Design rule registration system X: Design rule generation device 2: Terminal 11: Calculation unit 12: Operation display unit 13: Communication control unit 14: Storage unit p1: Design rule registration program p2: Web server program d1: Rule setting information d2: Design rule information d11: Rule condition part setting information d12: Rule conclusion part setting information I1: Matching condition setting information I2: Negative condition setting information I3: Lower limit condition setting information I4: Upper limit condition setting information S1 , S2,...: Processing procedure identification codes

Claims (6)

A production rule generation device for generating a production rule used in a rule-based system based on predetermined input information,
For the condition part of the production rule, at least condition part setting information comprising a combination of rule identification information for uniquely identifying each production rule and a condition value associated with each of a plurality of preset condition items, First condition part setting information for setting a matching condition, second condition part setting information for setting a mismatch condition, third condition part setting information for setting a numerical range lower limit condition, and fourth for setting a numerical range upper limit condition Condition part setting information input means for distinguishing and inputting four types of condition part setting information;
Conclusion part setting information input means for inputting conclusion part setting information consisting of a combination of the rule identification information and a conclusion value associated with each of a plurality of preset conclusion items for the conclusion part of the production rule; ,
One production rule in which the condition part and the conclusion part are set based on the combination of the four types of condition part setting information and the conclusion part setting information having the same rule identification information. Automatic production rule generation means for automatically generating and storing in a storage means,
The production rule automatic generation means is
In the first condition part setting information and the second condition part setting information having the same rule identification information, the condition value exists in the first condition part setting information and the second condition part setting information The condition item in which the condition value does not exist in the information for use is discriminated, and the condition of the production rule is that the condition item matches the content corresponding to the condition value of the first condition part setting information. A matching condition part setting means for setting the condition part;
The condition item in which the condition value exists in both the first condition part setting information and the second condition part setting information having the same rule identification information is determined, and the first condition is related to the condition item. Mismatch condition part setting means for setting the condition part of the production rule having a conformance condition that the content does not match the content corresponding to the condition value of the part setting information;
Regarding the condition item in which the condition value exists in at least one of the third condition part setting information and the fourth condition part setting information having the same rule identification information, the third condition part setting information and the Numerical range condition part setting means for setting the condition part of the production rule with a numerical range having a numerical value corresponding to the condition value in the fourth condition part setting information as a threshold;
The conclusion item in which the condition value exists in the conclusion part setting information having the same rule identification information is discriminated, and the content corresponding to the conclusion value of the conclusion part setting information is concluded with respect to the conclusion item Conclusion part setting means for setting the conclusion part of the production rule;
A production rule generation device comprising: The condition part setting information input means displays information on a plurality of the condition items on a predetermined display means, and each of the first condition part setting information to the fourth condition part setting information is individually set to a predetermined value. According to the input operation through the operation unit, the condition value corresponding to each of the condition items and the rule identification information are input in association with each other,
The conclusion part setting information input means displays a plurality of conclusion item information on the display means, and the conclusion value and the rule corresponding to each conclusion item according to an input operation through the operation part The production rule generation device according to claim 1, wherein the identification information is input in association with each other. The condition part setting information input means displays information on a plurality of the condition items on a predetermined display means, and according to an input operation through a predetermined operation part, the condition value corresponding to each of the condition items and The rule identification information is input in association with the condition distinguishing information for distinguishing between the first condition part setting information and the fourth condition part setting information.
The conclusion part setting information input means displays a plurality of conclusion item information on the display means, and the conclusion value and the rule corresponding to each conclusion item according to an input operation through the operation part The production rule generation device according to claim 1, wherein the identification information is input in association with each other. The condition part setting information input means reads the information from the predetermined storage means or receives the information from the external device through the predetermined communication means, thereby setting the first condition part setting information to the fourth condition part setting information. Enter each piece of information individually,
The conclusion part setting information input means inputs the conclusion part setting information by reading information from a predetermined storage means or receiving information from an external device through a predetermined communication means. The production rule generator described. The condition part setting information input means is associated with one of the information of the plurality of condition items by reading information from a predetermined storage means or receiving information from an external device through a predetermined communication means. A combination of the condition value, the rule identification information, and condition distinguishing information for distinguishing between the first condition part setting information and the fourth condition part setting information,
The conclusion part setting information input means inputs the conclusion part setting information by reading information from a predetermined storage means or receiving information from an external device through a predetermined communication means. The production rule generator described. A production rule generation program for causing a computer to execute a process for generating a production rule used in a rule-based system based on predetermined input information,
Computer
For the condition part of the production rule, at least condition part setting information comprising a combination of rule identification information for uniquely identifying each production rule and a condition value associated with each of a plurality of preset condition items, First condition part setting information for setting a matching condition, second condition part setting information for setting a mismatch condition, third condition part setting information for setting a numerical range lower limit condition, and fourth for setting a numerical range upper limit condition Condition part setting information input processing for input by distinguishing four types of condition part setting information;
Conclusion part setting information input processing for inputting conclusion part setting information consisting of a combination of the rule identification information and a conclusion value associated with each of a plurality of preset conclusion items for the conclusion part of the production rule; ,
One production rule in which the condition part and the conclusion part are set based on the combination of the four types of condition part setting information and the conclusion part setting information having the same rule identification information. A program that automatically generates production rules and stores them in the storage means.
The production rule automatic generation process is
In the first condition part setting information and the second condition part setting information having the same rule identification information, the condition value exists in the first condition part setting information and the second condition part setting information The condition item in which the condition value does not exist in the information for use is discriminated, and the condition of the production rule is that the condition item matches the content corresponding to the condition value of the first condition part setting information. A matching condition part setting process for setting a condition part,
The condition item in which the condition value exists in both the first condition part setting information and the second condition part setting information having the same rule identification information is determined, and the first condition is related to the condition item. A non-matching condition part setting process for setting the condition part of the production rule having a matching condition that does not match the content corresponding to the condition value of the part setting information;
Regarding the condition item in which the condition value exists in at least one of the third condition part setting information and the fourth condition part setting information having the same rule identification information, the third condition part setting information and the A numerical range condition part setting process for setting the condition part of the production rule with a numerical range having a numerical value corresponding to the condition value in the fourth condition part setting information as a threshold;
The conclusion item in which the condition value exists in the conclusion part setting information having the same rule identification information is discriminated, and the content corresponding to the conclusion value of the conclusion part setting information is concluded with respect to the conclusion item Conclusion part setting processing for setting the conclusion part of the production rule;
A production rule generator comprising:

Images