JPH0488428A - Approximate inference device - Google Patents

Abstract

PURPOSE:To shorten an inferring time by applying input data to a relation stored in a storage means and calculating the possibility of a conclusion, and calculating and displaying static information amounts or articulation of each event by conclusions. CONSTITUTION:The device consists of a knowledge storage device 11, a knowledge composition device 12, an after-composition knowledge storage device 13, an event value input device 14, an consistency arithmetic unit 15, a dynamic information amount arithmetic unit 16, a possibility arithmetic unit 17, a possibility display device 18, a static information amount arithmetic unit 19, an articulation arithmetic unit 20, an articulation storage device 21, an articulation addition device 22, an articulation display device 23, a static information amount display device 31, a display command device 32, and an event removal device 33. Then expert's knowledge showing the relation between events and conclusions is stored previously and the input data is applied to this knowledge to calculate the possibility of a conclusion by inference; and the articulation of each event regarding the possibility of each obtained conclusion is calculated and events which are small in static information amount or articulation among events regarding one conclusion are selectively deleted. Consequently, the processing time is shortened and the reliability of the conclusion is increased.

Description

[Detailed description of the invention] Expert knowledge (rules) expressing the relationship between summary events and conclusions of the invention is memorized in advance, and inferences are made by applying input data to this knowledge to determine the possibility of a conclusion. In an approximate inference device that calculates the clarity of each event related to the possibility of each conclusion obtained, selectively deletes those with static information content or low clarity among events related to the - conclusion. do. This allows rules to be organized, reducing processing time and increasing the reliability of the conclusions obtained.

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an approximate inference device that performs inference using expert knowledge representing the relationship between an event and a conclusion.

Prior Art Approximate inference is known in which the inference result is modified or changed depending on the amount of information about the event used to derive the inference result (for example, ZhangHongmin
AN EXPERT SYSTEM WITH)I
INKJNG IN] MAGES'', Prepri
nts of 5econd IPsA Congress
ss.

Tokyo, Nouly 20-25. 198
7. P, 765).

The second approximate inference method calculates the amount of information for each event (i.e., information discrimination ability of the event) using the membership function given for each conclusion for the event, and calculates the inference result (i.e., the probability that the conclusion holds true). ) by modifying or changing the amount of information of the event used to draw the conclusion (by multiplying the probability and the amount of information), it attempts to improve the discriminative ability of the inference result.

However, in such conventional devices, all useful events and non-useful events for drawing a conclusion are incorporated into the rules. For this reason, it is necessary to input a large amount of event data in actual approximate inference, which takes a long time to infer. This is a major disadvantage in applications such as fault diagnosis, which require accurate conclusions within a short period of time.

Summary of the Invention Purpose of the Invention An object of the present invention is to provide an approximate inference device that can organize rules (knowledge representing a relationship between a conclusion and an event) by deleting unuseful events from the rules.

Structure of the Invention 2 Actions and Effects The approximate inference device according to the present invention includes a storage means for storing the relationship between an event and a conclusion in advance, and a possibility of a conclusion by applying input data to the relationship stored in the storage means. approximate inference means for calculating the amount of static information or clarity of each event for each conclusion using the relationships stored in the storage means; and a display means for displaying the amount or clarity of static information.

The approximate inference device according to the present invention further includes event removal input means for inputting that a predetermined event should be removed from the relationship regarding the conclusion displayed on the display means.

In the approximate inference device according to the present invention, the probability of the conclusion is recalculated by the approximate inference means based on the relationship between the event and the conclusion after the event input from the event removal input means is removed from the relationship. .

Furthermore, in the approximate inference device according to the present invention, the static information amount or clarity is reconstructed by the calculation means using the relationship between the event and the conclusion after the event input from the event removal input means is removed from the relationship. Calculated.

The approximate inference device according to the present invention includes a storage means 2 that stores the relationship between an event and a conclusion in advance, and calculates the static information amount or clarity of each event for each conclusion using the relationship stored in the storage means. a calculation means for displaying the amount or clarity of static information calculated by the calculation means with respect to a predetermined conclusion; a predetermined event to be removed from the relationship with respect to the conclusion displayed on the display means; and approximate inference means for calculating the probability of a conclusion by applying the input data to the relationship between the event and the conclusion after removing the event input from the removal input means from the relationship. .

Then, the static information amount or clarity is recalculated by the calculation means using the relationship between the event and the conclusion after the event input from the removal input means is removed from the relationship.

According to the present invention, the static information amount or clarity of each of the plurality of events related to the - conclusion is displayed on the display means.

Looking at this display, the operator can input, using the removal input means, that an event with a static information content or low clarity, that is, an event that is not considered to be very useful, should be deleted. In this way, the knowledge (rules) representing the relationship between conclusions and events is organized, and a more concise and effective relationship between conclusions and events can be obtained.

Using knowledge representing the relationship between conclusions and events organized in this way, approximate inference to obtain the possibility of a conclusion and calculation of static information amount or clarity are performed again.

As described above, according to the present invention, it is possible to judge whether or not each event in the relationship (rule) between a previously set (designed) conclusion and events is useful, and therefore it is possible to evaluate an approximate inference device. becomes. Unuseful events are removed and a more effective and concise organized conclusion and relationship between events is obtained, making it possible to perform sufficiently practical approximate inference using less event input data, reducing inference time. be able to. If there is a large amount of static information or a large number of events with low clarity, the possibility of a conclusion obtained will be unclear, but according to this invention, since the above-mentioned events are removed, the possibility of a highly reliable conclusion is reduced. Obtainable.

Description of Embodiments (1) Overall configuration of approximate inference device FIG. 1 shows an example of the overall configuration of the approximate inference device. The approximate inference device includes a knowledge storage device 11. Knowledge synthesis device 1
2. Post-synthesis knowledge storage device 13. Event value input device 14. Fitness calculation device 15. Dynamic information amount calculation device 16. Possibility calculation device 17. Possibility display device 18. Static information amount calculation device 19. Clarity calculation device 20. Clarity storage 21. Clarity adder 22. Clarity display device 23, static information amount display device 31. It is composed of a display command device 32 and an event removal device 33.

These devices will be explained in detail below.

(2) Knowledge Storage Device The knowledge storage device 11 is a device that stores knowledge input by experts and the like in a format that shows the relationship between events and conclusions. This device can store the knowledge of multiple experts.

Two experts exl stored in the knowledge storage device 11
, ex2's knowledge is shown below in the form of a rule.

Expert exl: if 20≦fl≦60. 0≦f2≦40゜・,
a ≦r1≦b , -then clil
11...(1) if 40≦fl≦80. 60≦f2≦100゜-
, a , ≦f'i≦b i2, - then c2
...(2) Emperor f..., aij≦r1≦bij
, ・-then cj...(3) Expert ex2: if 30≦fl≦50.10≦f2≦30°-, a
, ≦f1≦b11゜then cl...(4) if 50≦f1≦70. 70≦12≦90゜・
..., a ≦fi≦bi2 '...then c
2...(5) if-,a,-≦N≦b,,,-then cjIJ
1 piece...(8) [1, r2. fi is an event, and these are respectively event 1. Event 2. Sometimes called event i. event index i
takes a value in the range of 1 to m. cl, c2. cj is the conclusion, and these are respectively concluded 1. Conclusion 2゜It is sometimes called conclusion j. The subscript j of the conclusion takes a value in the range of 1 to n.

Also, a i expressed as al, ≦fj≦b1j
j.

1 piece bo, are called the minimum value and maximum value, respectively.

]J The above rules are expressed in the form of a table for each expert as follows.

1st table expert exl 2nd table expert ex2 (3) Knowledge synthesizer The knowledge synthesizer 12 is a device that synthesizes the knowledge of multiple experts stored in the knowledge storage device tll and combines it into one piece of knowledge. be.

There are various ways to synthesize knowledge, but here we calculate the average value and standard deviation of multiple experts for the maximum and minimum values of each event involved in each conclusion.

The knowledge synthesis process will be described below, taking as an example the knowledge of the two experts described above that leads to the conclusion c1 from the event r1.

From the above rules (Equations (1) and (4)), event 1
By drawing the rules for obtaining conclusion 1 (cl) from (fl), it is expressed as follows.

Expert ext: if 20≦f1≦60 th
en cl-(7) Expert ex2: 1f80≦f
1≦50 then cl=−(8) The average value m of the minimum values and the average value ln m of the maximum values are calculated.

aX 20+ 80 -25 ... (9) 11
'l, g i.e.] n, 2 Table 3 The standard deviation σ of the minimum value and the standard 1n deviation σ of the maximum value are calculated.

The above-mentioned rules (
For equations (1) to (8), if all the minimum and maximum values of each event involved in each conclusion are examined, a table as shown below is obtained.

(Left below) Generally, in approximate reasoning, a membership function is given to an event. Here, as an example, a method for determining membership functions using Gaussian distribution using expert knowledge synthesized as described above will be described.

Average value m of minimum values, average value m of maximum values.

1111n
Using the standard deviation σ of the minimum value of l1aX, and the standard deviation in σ of the maximum value, the membership function is expressed as laX by the following equation.

...(13) however.

X: Value of input data to the event Φ(X) Degree to which the two-person data X fits the event (degree of fit) (iauss(x): Value of the Gaussian distribution at the input X.

Figure 2 shows an example of a Gaussian distribution. Only the left half of this Gaussian distribution is used to create the membership function. The position of X at Φ(X)-0,5 is m,
or m determined by ■1 n
IaX, the slope is determined by σ, or σ111laX.

As an example, the membership function for determining the conclusion c1 from the event fl is created as shown in FIGS. 3a to 3c using values calculated by equations (9) to (12). In this case, equation (18) becomes as follows.

...(14) Figure 3a is the first right-hand side of equation (18) or equation (14).
Figure 3b shows the second term on the right side of equation (13) or equation (14), and Figure 3C shows the result of subtracting the second term from the first term above, that is, equation (13) or equation (14). (14) is 4
represents the membership function to be represented.

Each event fl, r2. created based on the synthesized knowledge shown in Table 3. rN:: -) LN Te, conclusion c
l, c2. Examples of membership functions for determining cj are shown in FIGS. 4a, 4b and 4c.

(4) Post-synthesis knowledge storage device The post-synthesis knowledge storage device 13 stores the average value and standard deviation calculated by the knowledge synthesis device 12 in the format shown in Table 3. Since it is not necessary to synthesize knowledge every time an inference is made, the results calculated in advance in this way are memorized. By reading out and using the values in the storage device 13 each time an inference is made, it is possible to speed up the inference process.

(5) Event value input device The event value input device 14 is a device that reads input data input for each event from a keyboard, communication interface device, memory, file, etc. The input data is given to the fitness computing device 15, and information as to whether data for each event has been input is given to the clarity addition device 22.

The event value may be given not only as a definite value but also as a linguistic value or a membership function.

(8) Fitness calculation device The fitness calculation device 15 calculates the fitness of the data input from the event value input device 14 for each membership function (or conclusion). Specifically, if the input data is a definite value, the goodness of fit can be calculated by substituting the input data as the variable X on the right side of equation (13).
It is required as. Of course, such an arithmetic expression does not necessarily have to be used. If the input data is a linguistic value or a membership function, the degree of fitness may be calculated using MIN-MAX, for example.

(7) Dynamic information calculation device and static information calculation device event f
The event value of event r2 (input data) is xl, and the event value of event r2 is x2. Let the event value of event fi be xi. These data are input from the event value input device 14.

As shown in Figures 5a and 5b, each degree of fitness is "11".
'm12''lj''21'm22'
``2j is defined as 0 as follows.

11 'Fitness of the input data xi to the membership function of the conclusion C1' 12 'Fitness of the input data x1 to the membership function of the conclusion C2 jrllj: goodness of fit of the input data x1 to the membership function of the conclusion cj m21. Fit of input data x2 to the membership function of conclusion C1 ``22 - Fit of input data x2 to the membership function of conclusion C2 m2j: Fit of input data x2 to the membership function of conclusion ej Generally, as shown in Fig. 5C. As shown, the goodness of fit m1. is the goodness of fit of the input data (event value) xi to the membership function of the conclusion cj.

These fitness values are calculated based on the input data xi, x2. It is calculated by the fitness calculating device 15 when xi is given.

Let us now consider the concept of fuzzy entropy.

Fuzzy entropy Ef when human power xi is given
Define i as follows.

Eri −, '4 f (m, / M, ) log
(m, /Mi) 1...(15) Here M, -Σ m...
(18) 1 + - 113 The second fuzzy entropy is a type of index of information discrimination ability. When input data xi is given, the value is small enough that the conclusion can be clearly identified, and the value is small enough that the conclusion can only be vaguely identified. It becomes a large value.

The range of values that the fuzzy entropy Ef can take is as shown below.

0≦Ef≦log(n) n; number of conclusions on the event Next, using this fuzzy entropy Eft, the dynamic information amount Iff when input data xi is given, (
Find xi). Here, the amount of dynamic information 1fiD(xi)
is the ability to identify events in order to establish a conclusion when making inferences.

Therefore, the amount of dynamic information about the event number 1fia(xi)
is defined as the maximum fuzzy entropy minus the fuzzy entropy Eft when input data xi is given.

Ifin(xi)-1Cg(n)+,)? -i(mij
/M, ) log (mij/M, ) 1...
(17) The dynamic information amount of the event f1 to which no input data xi is given is zero.

The dynamic information amount calculating device 16 uses the fitness obtained by the fitness calculating device 15 to calculate the dynamic information amount for each event according to equation (17).

The amount of dynamic information depends on the input data xi as described above.

On the other hand, the amount of static information does not depend on the input data, and the static information of the entire event is calculated by subtracting the average fuzzy entropy within the range of the event from the maximum fuzzy entropy. Quantity. The amount of static information about event i is determined by the following formula% m ij (x 'k): The degree of fitness of conclusion cj to the membership function of input data X2 for event ri When calculating the amount of static information, the fourth c. As shown in the figure,
The range width Xll to xjp of the event fi is expressed at intervals of δ (p-1
). (1/p) [Σ
81. ] represents the operation of calculating the fuzzy entropy for each X l k and finding their average.

As can be seen from equations (18) and (19), the greater the overlap between the event membership functions.

The amount of static information of an event is small, and the smaller the overlap between the membership functions of an event, the larger the amount of static information of an event.

In other words, the amount of static information indicates the ability of the event member function to identify the conclusion.

The static information amount calculating device 19 calculates and stores the amount of static information for each event from the membership function obtained from the synthesized knowledge according to the above-mentioned equations (18) and (19). do. Since the static information amount does not depend on input data, it only needs to be calculated once.

(8) Possibility calculation device For each conclusion, calculates the amount of information of events such that the sum of the amount of information of events involved in that conclusion is 1 and the relative strength of the amount of information of those events does not change. . This calculated amount of information is called a weight.

For example, if the above-mentioned dynamic information amount is used, the weight vl for each event fi will be as follows.

vi= If (xi)/1 child 1fi, (xi)
=-(20) Next, calculate the product of these weights and the goodness of fit.

The total sum for each conclusion is calculated as the probability Pr of each conclusion cj.

For example, in the above example, Pr, -Σ (vj x m ij) ・
-(21)J 1+1 The possibility calculation device 17 performs the above-mentioned calculation to calculate the probability for each conclusion.

(9) Possibility Display Device This possibility display device 18 displays the possibilities calculated by the possibility calculation device 17 for each conclusion. The possibility may be displayed for all conclusions, or one or more highly probable conclusions may be displayed. Also, the possibility may be transmitted to another device through one communication, or the possibility may be stored in a memory or a file.

(10) Clarity calculation device The clarity calculation device W20 is a device that calculates the clarity of each event for each conclusion. Here, the clarity of each event for each conclusion indicates the relative discernment ability of each event when evaluating the possibility of a certain conclusion. Therefore, this clarity makes it possible to compare the discriminative ability of multiple events in order to establish a certain conclusion, and to understand which event has a high discriminative ability (contains a large amount of information). Ru.

The clarity of each event for each conclusion is determined by the following formula.

however.

A1. -Σ m,, (xi) IJ *-+ IJ k That is... (23) If the membership function of conclusion cj exists in event f i, then the membership function of conclusion cj in event fI. If the function does not exist, A,, -0 J The denominator of equation (22), which represents the clarity of each event for each conclusion, is the event rl for which the membership function of conclusion ej exists.
The total sum of the static information amount 1fis is calculated only for .

Since clarity is obtained by normalizing the amount of static information using equation (22), it is easy to intuitively understand the relative discrimination ability.

In this way, clarity calculation bag! In step 20, the clarity of each event is calculated for each conclusion.

(11) Clarity Storage Device The clarity storage device 21 is a device that stores the clarity of each event for each conclusion calculated by the clarity calculation device 20. Clarity calculations do not need to be performed every time an inference is made. Therefore, the clarity calculated when knowledge is synthesized is stored in the clarity storage device 21.

The value stored in the clarity storage device 21 is read out every time an inference is made. This makes it possible to speed up one inference process.

(12) Clarity Addition Device The clarity addition device 22 is a device that calculates the clarity of an event to which data is actually input for each conclusion. here,
for the inferences that are actually made.

Take the summation for each conclusion of the clarity of the events for which data has been entered. The sum of this clarity indicates the clarity of the inference result. It can be said that the higher the clarity, the greater the amount of information for deriving the inference result (conclusion). Therefore, clarity can be used as an indicator to judge the reliability of the inference result itself.

The clarity for the inference results (that is, for each conclusion) is calculated according to the following formula.

(1, -1 壬、Cρ8. ...(2
4) However, in equation (24), the clarity cp of the event ri for which no input data xi is given is treated as 0. That is, the addition of equation (24) is performed in one row only for the clarity CL of the event to which input data xj is given.

Clarity of inference results cI! What is the possible range of ?

ko0.0≦C1,≦1.0 ko. In other words, if you make an inference by inputting data about all the events that can be used to draw a certain conclusion within the knowledge given before making the inference,
The clarity of the conclusion will be 1.0. Also, only some of the events that can be used to draw a certain conclusion.

If the data is entered, clarity will be a value between 0.0 and 1.0. At this time, it can be said that if a large number of events with high clarity are used among the events that can be used, the clarity of the conclusion will also be increased, and a highly reliable inference result can be obtained.

(13) Clarity Display Device The clarity display device 23 is a device that displays the clarity of the inference result (-the possibility mentioned above as an example) calculated by the clarity addition device 22. The clarity may be displayed along with the inference results, or the clarity may be transmitted to another device or stored in memory or a file.

This clarity display displays all conclusions of the inference results. Therefore, if there are multiple conclusions, the clarity corresponding to each conclusion is displayed.

In this way, every time data is entered.

By calculating the information amount of the event to which the input data belongs and displaying the clarity of the inference result, the user can judge the reliability of the inference result.

(14) Static information amount display device and display command device As mentioned above, the static information amount for each event is calculated by the calculation device 19, and the results are presented in the form of a table for each conclusion, as shown in Figure 6. and is stored in the arithmetic unit 19 or the static information amount display device 31. In the table of FIG. 6, the order is the order of the magnitude of the amount of static information.

When the number j or code specifying a conclusion is input from the display command device 32, the static information amount display device 31 reads out the amount of static information related to that conclusion from the table shown in FIG. A total of 10 static information amounts, 5 in descending order and 5 in descending order, are displayed. When the operator looks at these displays and determines that there is a large variation in the amount of static information between events, and wants to see the amount of static information for all events, he inputs a display command for all events from the display command device 32. do. The display device 31 then displays the amount of static information of all events related to the given conclusion.

In this way, the two workers can judge the amount of static information of each event regarding the desired conclusion, that is, whether each event is useful (contributes) to the conclusion, by looking at the display.

As mentioned above, the clarity of each event per conclusion, c, p,.

J can be found by normalizing the amount of static information, so
This clarity C111 is stored in a form similar to the table in Figure 6, and the clarity cn is used instead of the static information amount.
, , may be displayed on the display device 31.

(15) Event removal device The event removal device 33 is a static information display bag! This is for inputting J events that one operator has determined should be deleted by looking at the static information amount (or clarity C1) of each event displayed in 81.

For example, an operator compares five events with a large amount of static information displayed on the display device 31 and five events with a small amount of static information, and determines whether the difference in the amount of static information is very large (
For example, if there is an event whose static information amount is twice or more than the small static information amount, an input is input from the device 33 that the event with a small static information amount of one should be deleted.

(1B) Recalculation of probability, clarity, etc. When one or more events to be deleted for a certain conclusion are input from the event removal device 33, recalculation of probability, clarity, etc. is performed. That is, the data representing the events to be deleted are sent on the one hand to the goodness-of-fit calculation device 15;
In the fitness calculation in step 5, the calculation of the fitness of the membership function regarding the above conclusion of the event determined to be deleted is omitted. So, possibility calculation device 1
Input data of events determined to be deleted will not be included in the possibility calculation in step 7.

Deleting an event means removing the event from the rules representing the knowledge described above. In this way, a newly calculated possibility is displayed on the display device 18 for the corresponding conclusion according to the rule from which the event inputted from the device 33 is removed.

This allows one worker to judge whether the deletion of the event was appropriate or not.

Similarly, the data representing the event to be deleted inputted from the device 33 is also given to the static information amount calculation device 19,
The amount of static information is recalculated according to the rule that removes the events that should be deleted.

Also, clarity calculations and clarity additions are not performed again. The clarity of each conclusion recalculated in this way is displayed on the display device 23, and the amount of static information is displayed on the display device 31.

Displaying the amount of static information, deleting and inputting events, and recalculating according to rules excluding deleted events are repeated for each conclusion while changing the conclusion number. Finally, the rules for all conclusions are organized into a more concise form. This process can be briefly summarized as shown in FIG.

It goes without saying that each of the devices 11 to 23 and 31 to 33 described above can be realized by a computer including a memory and a display device. For example, the various arithmetic units 15, 16, 17, 19, 20, and 22 of the intellectual and technological synthesis device 12 are suitably realized by a CPU that operates according to a program.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an example of the overall configuration of an approximate inference device. FIG. 2 is a graph showing a Gaussian distribution. Figures 3a to 3c are graphs showing how membership functions are formed. Figures 4a, 4b and 4c are graphs showing the membership functions obtained for each event. FIGS. 5a, 5b, and 5c are graphs showing how the goodness of fit is determined. FIG. 6 shows a static information amount table. FIG. 7 is a flow chart showing event deletion processing. 32...Display command device. 33...Event removal device.

Claims (1)

[Scope of Claims] (1) Storage means for storing the relationship between events and conclusions in advance, and approximate inference means for calculating the possibility of a conclusion by applying input data to the relationship stored in the storage means. , calculation means for calculating the static information amount or clarity of each event for each conclusion using the relationships stored in the storage means;
and display means for displaying the amount or clarity of static information calculated by the calculation means with respect to the predetermined conclusion. Approximate reasoning device with (2) The approximate inference device according to claim (1), further comprising means for inputting that a predetermined event should be removed from the relationship regarding the conclusion displayed on the display means. (3) A claim further comprising means for controlling the approximate inference means to calculate the probability of a conclusion based on the relationship between the event and the conclusion after removing the event input from the removal input means from the relationship. The approximate inference device according to item (2). (4) Further comprising means for controlling the calculation means to calculate static information amount or clarity using the relationship between the event and the conclusion after removing the event input from the removal input means from the relationship. The approximate inference device according to claim (2). (5) A goodness-of-fit calculation means in which the approximate inference means converts the input data into goodness of fit using a membership function expressed by the above relationship, and this goodness of fit is used to calculate the amount of dynamic information for each event. 2. The approximate inference device according to claim 1, comprising: a dynamic information amount calculation means; and a possibility calculation means for calculating the possibility of a conclusion using the fitness degree and the dynamic information amount. (8) The approximate inference device according to claim (1), further comprising a display device that displays the possibility of the conclusion. (7) Calculate the clarity of each event for each conclusion using the amount of static information obtained from the calculation means, and use the calculated clarity or the clarity obtained from the calculation means, The approximate inference device according to claim 1, further comprising: adding means for calculating the clarity of each conclusion by adding up the clarity of events for which data is actually input. (8) The approximate inference device according to claim (7), further comprising: clarity display means for displaying the clarity of the inference result obtained from the clarity addition means. (9) A memory means for storing the relationship between events and conclusions in advance. Calculating means for calculating the static information amount or clarity of each event for each conclusion using the relationships stored in the storage means. Display means for displaying the static information amount or clarity calculated by the calculation means with respect to a predetermined conclusion. With respect to the conclusion displayed on the display means, a means for inputting that a predetermined event should be removed from the relationship, and a relationship between the event and the conclusion after the event input from the removal input means is removed from the relationship. Approximate inference means for calculating the probability of a conclusion by applying input data to . (10) Further comprising means for controlling the calculation means to calculate static information amount or clarity using the relationship between the event and the conclusion after removing the event input from the removal input means from the relationship. The approximate inference device according to claim (9).