JPH0793394A - Design equipment - Google Patents

Abstract

(57) [Summary] [Purpose] The objective is to provide a design device that has the same impression as the initial candidate and can infer the next candidate design with a different atmosphere in less time. [Structure] A knowledge database 2, a keyboard 1 for inputting evaluation terms 9, an inference means 3 for inferring, for each item, a category having the maximum relationship with the evaluation terms 9 using the knowledge database 2. In a design device provided with a CRT 5 that displays the inferred design element 11, a next candidate inference means 4 for inferring and extracting a next candidate in place of the inferred design element is provided, and the next candidate inference means 4 is for each item. When the category of the next candidate is selected, the strength of the relationship between the inferred category and the category of the next candidate with respect to the input evaluation term is taken for each item, and the difference in the strength is set as a reference difference set in advance. The inferred category is changed to the next candidate category for items that are within the range.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a design device for confirming a human image and displaying a design on a display device such as a CRT.

[0002]

2. Description of the Related Art Conventionally, as a design device of this type,
For example, there is one as shown in FIG. 11 disclosed in Japanese Patent Laid-Open No. 4-278672.

This design apparatus includes an input means 1 such as a keyboard, a knowledge database 22, and an inference means 3.
And a correction inference means 4 and a display means 5 such as a CRT. The knowledge database 22 includes an evaluation term 9, an adjective knowledge 10, a vehicle design element 11, and an image knowledge 3.
2 and.

The vehicle design element 11 constituting the design object is first classified into a plurality of items, and each item is further classified into a plurality of categories. Then, the degree to which each item affects each of the evaluation terms 9 representing the design image is expressed by a partial correlation coefficient (numerical value) for each item, and similarly, for each of the evaluation terms 9, The degree to which each of a plurality of categories affects each item is represented by a partial regression coefficient (numerical value) for each category, and is stored as a database as image knowledge 32.

First, when a user operates a keyboard to input an evaluation term 9 representing a design image that the user has, first, the inference means 3 uses the knowledge database 22 to make a maximum deviation with respect to the evaluation term 9. A category having a regression coefficient is inferred and extracted as an initial candidate for each item, and the initial candidate is displayed on the display means 5 as an image.

When the user who sees the initial candidate displayed in the image feels that the image does not match his / her own image,
Operate the keyboard to instruct execution of the next candidate inference. Then, this time, the modified inference means 4 uses the knowledge database 22 to infer the next candidate. At this time, the correction inference means 4
The category having the smallest partial regression coefficient in the item having the smallest partial correlation coefficient in the image knowledge 32 or in the item having the partial correlation coefficient less than or equal to a preset reference value is inferred and extracted as the next candidate. . In response to a query from the CRT as to whether or not the next candidate image is satisfied, when the user again instructs execution of the next candidate inference, this time, one of the partial regression coefficients of the item having the second smallest partial correlation coefficient is detected. Infer and extract the smallest category as the next candidate,
The extracted next candidate is displayed as an image instead of a part of the initial candidate already displayed as an image. Then, this process is repeated to correct the image of the user.

[0007]

However, in this design device, as described above, when the user infers the next candidate when he feels that the image does not match his / her own image, the initial candidate already displayed as an image is The category having the smallest partial correlation coefficient in the item having the smallest partial correlation coefficient or the preset reference value or less is sequentially inferred and extracted as the next candidate. In other words, the next candidate design that contains changes that have little impact on the initial candidate design is sequentially inferred and extracted, and the inference is repeated many times to gradually adjust to the user's image, so the following problems remain. It was

That is, it is difficult to know where the next candidate displayed with respect to the design which has already been displayed as an image has changed, and the atmosphere changes little. Also, for example, for one input evaluation term "sporty",
Different inference results cannot be obtained as if there are multiple atmospheres. In other words, it is impossible to make multiple inferences that there are A and B designs that have the same impression of a sporty interior but completely different moods. Furthermore, since there is little change in the next candidate, the next candidate inference is repeated, and the problem that this repeated inference takes time remains.

The present invention has been made by paying attention to such a problem, and an object thereof is to provide a designing apparatus which can infer a next candidate design having the same impression as that of an initial candidate and a different atmosphere in a short time. And

[0010]

In order to solve the above problems, the invention according to claim 1 relates to a relationship between a plurality of items and design elements classified into a plurality of categories for each item and design evaluation terms. A knowledge database that stores information, input means for inputting the evaluation terms, and a category having the maximum relationship with the evaluation terms by using the knowledge database based on the input evaluation terms. In a design device including an inference means for inferring each of the inferred design elements and a display means for displaying the inferred design elements in combination, the next candidate inference means for inferring and extracting the next candidate design element replacing the inferred design element. When the next candidate category is selected for each item, the next candidate inference means selects the next candidate category and infers the inferred value for each item. Taking the difference in the strength of the relationship between the Gory and the next candidate category with respect to the input evaluation terms, and regarding the item whose strength difference is within a preset reference difference, the inferred category is the category of the next candidate. It is characterized by changing to.

The invention according to claim 2 is the design apparatus according to claim 1, wherein the next candidate inference means presets the strength of the relationship that the item has with respect to the input evaluation term. It is characterized in that the category is changed for items whose difference in strength of the relationship is less than or equal to the standard and within the standard difference.

The invention according to claim 3 is the design apparatus according to claim 1, wherein when the change to the next candidate is made over a plurality of items, the next candidate inference means is the next candidate between the items. It is characterized in that all the combinations of the categories including the categories are extracted, and the change is made for each combination.

The invention according to claim 4 is the design apparatus according to claim 1, wherein the next candidate inference means combines the categories with the evaluation terms when the change to the next candidate involves a plurality of items. It is characterized in that the change is performed by selecting a category of a matching combination between the items by using the knowledge about.

[0014]

According to the first aspect of the present invention, when the evaluation term is input, the inference means infers the category having the maximum relationship with respect to each item, and the inferred design element ( Category) is displayed in combination.
After that, when inferring and extracting the next candidate design element that replaces a part of this design element, the next candidate inference means selects the next candidate category for each item, and the inferred category and the next candidate are selected for each item. The difference between the strength of the relationship with the category and the input evaluation term is calculated, and the inferred category is changed to the category of the next candidate for an item whose strength difference is within a preset reference difference. Then, the display means displays the next candidate in place of a part of the design elements already displayed.

According to the second aspect of the invention, when inferring and extracting the next candidate category in place of a part of the design elements that have already been displayed, the next candidate inferring means determines the input evaluation term. The category is changed for an item in which the strength of the relationship of the items is equal to or less than a preset standard and the difference in the strength of the relationship is within the standard difference. Then, the display means displays the next candidate in place of a part of the design elements already displayed.

According to the third aspect of the present invention, when the change to the next candidate is made over a plurality of items, the next candidate inference means sets all combinations of categories including the category of the next candidate among the items. It is extracted and the change is made for each combination. Then, the display means displays the next candidate in place of a part of the design elements already displayed.

According to the invention described in claim 4, when the change to the next candidate is performed over a plurality of items, the next candidate inference means uses the knowledge about the combination of the categories with respect to the evaluation terms to make a difference between the items. The category of a suitable combination is selected and the change is performed. And the display means is
Display the next candidate in place of some of the design elements already displayed.

Thus, according to the first to fourth aspects of the invention, the next candidate inferred and extracted by the next candidate inference means has a great influence on the already displayed design element. Therefore, the impression is the same, but the mood is different, and moreover, since it is inferred rather than repeated, it takes no time to display the next candidate.

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A first embodiment of the present invention will be described with reference to FIGS. Since this embodiment has the same structure as that of the conventional example, the same members as those of the structure of the conventional example are designated by the same reference numerals in the drawings.

As shown in FIG. 1, the designing apparatus of the first embodiment comprises a keyboard 1 which is an input means, a computer 16 and a CRT 5 which is a display means.
The knowledge database 2 is stored in the memory of the computer 16. Further, the CPU of the computer 16 incorporates the inference means 3 and the next candidate inference means 4. Then, the knowledge database 2 further includes an evaluation term 9 for expressing a design image, an adjective knowledge 10, a design element 11 and an image knowledge 12 constituting a vehicle design.
It has and.

The evaluation term 9 is composed of an adjective word group as shown in the evaluation term column of FIG.

As shown in FIG. 2, the adjective knowledge 10 is the knowledge obtained by collecting the evaluation terms 9 and analyzing the result of the preliminary experiment to define the relationship between the evaluation terms. Is, for example, data obtained by dividing the evaluation term 9 into a plurality of factors by multivariate analysis such as factor analysis and obtaining as a factor loading (numerical value) how much each evaluation term 9 is related to each factor. . It can be said that the evaluation terms having a close numerical value are closely related and similar to each other when the factor load is particularly large. This adjective knowledge 10 is used to select the most similar evaluation term 9 when the input evaluation term is not among the evaluation terms 9 used in the sensory evaluation experiment described later when using the design device. It Also,
Apart from this, another method may be adopted in which another evaluation term close to the input evaluation term is easily selected. An example thereof will be described later with reference to FIGS. 9 and 10.

The design element 11 is, as shown in FIG.
First, it is classified into a plurality of items, and each item is further classified into a plurality of categories. That is, the items in FIG. 3 are focused on the number of meters provided on the instrument panel in the design element 11, the instrument panel and the meter cluster, the meter cluster and the center console, the thickness of the door, and the overhang of the center cluster. is there.

The items focusing on the number of meters are classified into five categories. That is, the first category with one large meter, the second category with one large meter and two small meters, the third category with one large meter and three small meters, the two large meters. 4th category with one and one small meter, large meter 2
They are categorized in the 5th category, which has two pieces and two small meters.

Items focusing on the instrument panel and the meter cluster are classified into two categories. That is, the instrument panel and the meter cluster are integrated into a first category and a separated second category.

Items focused on the meter cluster and the center console are classified into two categories. That is, the meter cluster and the center console are integrated into the first category and the separated second category.

Items focusing on the thickness of the door are classified into two categories. That is, the thick first door
It is categorized into a category and a second category with thin doors.

Items focused on the overhang of the center cluster are classified into two categories. That is, the center cluster is classified into the first category with overhang and the second category without overhang.

Items other than those shown in FIG. 3 are also classified into a plurality of categories.

The image knowledge 12 is, as shown in FIG.
It was obtained as a result of a sensory evaluation experiment. Specifically, the items of the evaluation term 9 and the design element 11 are selected, and the design element 11 corresponding to each item is selected.
Shown to an unspecified number of people, and the evaluation term 9 which is an adjective group
The feelings received from the design elements 11 for each of the above are collected, and the collected results are analyzed as partial regression coefficient (partial correlation coefficient) by multivariate analysis such as quantification theory class I or class II. is there.

Further, FIG. 4 shows a part of the image knowledge 12 of FIG. 3, but it is an image knowledge of an example of the evaluation term “sporty” not shown in FIG. Here, the item is described in the left column, and the category is described in the center.

Next, when the evaluation term 9 is input, the inference means 3 follows the basic flow chart shown in FIG. 5, and the category (initial candidate) in which the strength of the relationship from the knowledge database 2 to the evaluation term 9 is maximum. ) Is inferred and extracted, and the image is displayed on the CRT 5.

The next candidate inference means 4 operates the next candidate selection instruction on the keyboard 1 by the user, and in accordance with the next candidate inference flowchart shown in FIG.
Infering the design element 11 in which the difference in the strength of the relationship with the evaluation term input from is smaller than a preset standard,
A part of the initial candidates that have been extracted and already displayed as images are replaced with the next candidates and displayed on the CRT 5 as images.

Next, the operation of this design device will be described.

FIG. 5 is a basic flow chart of this apparatus. A description will be given along this flowchart. First, the design apparatus is started in step 101, and step 10
When the evaluation term 9 is input in step 2, the inference means 3 infers and extracts the category having the maximum partial regression coefficient for each design element from the knowledge database 2 in step 103. In step 104, the combination of the categories (initial Candidate) is CR
Displayed at T5. Then, in step 105, the user is inquired whether or not this initial candidate matches the image of the user (necessity of inference of the next candidate), and if the user does not match (necessary inference of the next candidate), the keyboard 1 by the user By the next candidate inference instruction, the process proceeds to step 106.

In step 106, the next candidate inference means 4 infers the next candidate (described later), and in step 107, the next candidate design in which a part of the initial candidates is replaced with the next candidate is CR.
Display at T5 and end at step 108.

On the other hand, if it is determined in step 105 that the initial candidate matches the image of the user (no need to infer the next candidate), the process ends in step 109.

If the evaluation term input in step 102 is not in the image knowledge 12 (FIG. 3), the evaluation term most closely related to the term is selected from the adjective knowledge 10 (FIG. 2) and used. If it is not in the adjective knowledge 10, the evaluation term of a different expression is re-input.

Next, the flow of the next candidate inference by the next candidate inference means 4 will be described.

FIG. 6 is a flowchart for inferring the next candidate design element 11 by the next candidate inference means 4 in step 106.

The next candidate inference means 4 uses the image knowledge 12 to infer the next candidate. Here, the case where the evaluation term 9 "sporty" is input is taken as a specific example, and the next candidate inference is performed. The flow will be described. The image knowledge 12 used for inference is shown in FIG. First, when the next candidate inference is started in step 201, in step 202,
The image knowledge 12 (FIG. 4) is referred to, and the category with the maximum partial regression coefficient is determined for each item (the category column in FIG. 4 is marked with an O, which is the first category, and is the initial candidate. Is). Next, for each item, calculate the numerical difference between the one with the largest partial regression coefficient and the one with the next largest partial regression coefficient (this is the difference from the number underlined in the category column in the figure, and the calculated value is Fill in the column of "difference between first category and second category" (column A) and store it). Then, in step 203, an item whose calculated numerical difference is within a preset reference difference (for example, 0.1) is selected (items marked with ♦ in the margin of the item column in the figure). In this example, the items "Instrument thickness" and "Meter cluster and cluster C" are selected. Then, in step 204, the total number of combinations of the first and second categories including the second category (marked with □ in FIG. 4) of the selected item is counted. That is, from FIG. 4 a) a combination of “medium” and “separated”,
The total number of combinations of the following candidates is three: b) a combination of “thick” and “integrated type 1”, and c) a combination of “thick” and “separated”. So, in step 205, all three of the above,
That is, a plurality of next candidate combinations are displayed on the CRT 5 at once or sequentially. Figure 7 shows the case of displaying on CRT5 at once.
It shows in (a). In the case of sequential display, the number of remaining next candidates is also displayed at the same time. This case is shown in FIG.
It should be noted that the CRT screen is temporarily returned to the initial candidate display before the sequential replacement with the next candidate image. Then, in step 206, the flow ends.

In selecting the next candidate in step 204, another method for selecting the next candidate, which is different from the above, is shown in FIG.
A method of selecting the next candidate in order from the item having the smallest numerical value in the column A and replacing it with the category of the initial candidate may be adopted. In this case, “meter cluster and cluster C” in FIG. 4 are selected in order.

Thus, according to the first embodiment, when the next candidate is inferred, the numerical difference between the largest partial regression coefficient and the second largest partial regression coefficient is within the reference difference of 0.1,
Select the next candidate category. That is, for the "sporty" evaluation term, a category having a strength close to the strength of the relationship of the initial candidate is selected as the next candidate. Unlike the conventional example, this selection method is performed regardless of the size of the partial correlation coefficient (see the partial correlation coefficient column in FIG. 4) of each item, and because it is inferred rather than repeated, it is input. For the evaluation term "sporty", the impression is the same as that of the initial candidate, but the atmosphere is different, and the next candidate can be displayed in a short time.

Next, a second embodiment of the present invention will be described with reference to FIGS.
Will be explained.

The second embodiment differs from the first embodiment in the method of selecting the next candidate when inferring the next candidate. for that reason,
In the configuration of the second embodiment, the image knowledge 12 is
It also has combination knowledge 12a regarding the combination of the design elements 11 with respect to the evaluation terms 9, as shown in FIG. This combination knowledge 12a is a database of knowledge about combinations of categories between items that designers and other experts consider common sense.
In FIG. 8, for example, in order to make the interior of a "warm" image, if the instrument shape is round,
Combination knowledge that the thickness of the instrument is preferably thick can be used for inference of the next candidate.

In the second embodiment, step 2 in FIG.
Combination knowledge 1 out of all combinations of the next candidates counted in 04
Use 2a to narrow down the next candidate category. That is, referring to the combination knowledge 12a (FIG. 8), the “instrument thickness” is medium and the “meter cluster and cluster C” are combination of separation with respect to the “sporty” evaluation term (marked with ♦ in FIG. 8). Since this is found to be good, this combination is selected as the next candidate. Then, this next candidate is displayed on the CRT 5.

In this way, according to the second embodiment, since the combination knowledge 12a is used in selecting the next candidate, the next candidate having the same impression as the initial candidate but a different mood with respect to the input evaluation term is selected. You can narrow down efficiently and display in less time.

Next, a third embodiment of the present invention will be described with reference to FIG.

The third embodiment is based on the inference of the next candidate.
The setting condition is different from that of the first embodiment, and the configuration is the same. Therefore, this difference will be described with reference to the example of FIG.

In the third embodiment, when inferring the next candidate, first, an item in which the numerical difference between the largest partial regression coefficient and the second largest partial regression coefficient is within the reference difference of 0.1 is selected. Then, among the items selected next, the partial correlation coefficient of the item is set to a preset reference value (for example, 0.
9) Select the following items. In this example,
The marked items, that is, the items "instrument thickness" and "meter cluster and cluster C" are selected, as in the case of the first embodiment.

In the subsequent flow, for example, the next candidate inference is performed in the same flow as in the case of the first embodiment.

Thus, according to the third embodiment, when the next candidate is inferred, the numerical difference between the one having the largest partial regression coefficient and the one having the next largest partial regression coefficient is an item whose reference difference is within 0.1.
Moreover, the item whose partial correlation coefficient is 0.9 or less is selected. In other words, by increasing the reference value of the partial correlation coefficient in this way, the next candidate having a strong relationship with the input evaluation term is selected, so that the next candidate having the same impression as the initial candidate but a different atmosphere can be selected. It can be reasoned and displayed quickly.

Next, a fourth embodiment of the present invention will be described with reference to FIG.

The fourth embodiment is based on the inference of the next candidate.
Based on the arrangement of the input evaluation terms, another evaluation term having a meaning close to the input evaluation terms is selected and the image knowledge 12 of the evaluation terms is used. That is, for example, when the evaluation term “for a running person” is input, FIG.
As shown in (a), when the next candidate inference is started in step 301, in step 302, as shown in FIG. 9 (b), evaluation of “many” or “sporty” with a close distance on the cloth placement. Choose terms and in step 303, using image knowledge 12 for each of these assessment terms,
For example, the next candidate is inferred by the same method as the inference of the first embodiment. And

At step 304, the inferred design element 11 is displayed on the CRT 5 as the next candidate, and at step 3
At 05, the flow ends.

As another inference method similar to the fourth embodiment, a fifth embodiment of the present invention will be described with reference to FIG.

The fifth embodiment is based on the inference of the next candidate.
Another evaluation term having a meaning close to the input evaluation term is selected from the input evaluation term cluster, and the image knowledge 12 of the evaluation term is used. That is,
For example, if you enter the evaluation term "for runners",
As shown in FIG. 10 (a), when the next candidate inference is started in step 401, in step 402, as shown in FIG. 10 (b), the term in the cluster of <sporty> is “spartan”, A term such as "sporty" is selected, and in step 403 the next candidate is inferred using the image knowledge 12 for each of these terms, for example, in the same manner as the inference of the first embodiment. Then, in step 404, the inferred design element 11 is displayed on the CRT 5 as the next candidate, and in step 405, the flow ends.

In this way, according to the fourth and fifth embodiments, the next candidate is inferred using the image knowledge 12 for each of the evaluation terms that have a similar meaning to the input evaluation term. Efficiently view the next candidate that matches the user's image on the CRT without trial and error, such as newly inputting another evaluation term in an attempt to display the next candidate other than the one displayed. You can

[0059]

As is apparent from the above description, according to the invention described in claim 1, when the next candidate category is selected for each item when the next candidate design is inferred, the initial candidate is selected for each item. The difference between the strength of the relationship between the input category and the category of the next candidate and the category of the next candidate is calculated, and the category of the initial candidate is set as the next candidate for the item whose difference in strength is within a preset reference difference. Change to category. And since this next candidate is displayed instead of the initial candidate, the impression is the same as the initial candidate,
You can infer next candidate designs with different atmospheres in less time.

According to the second aspect of the invention, when inferring the next candidate design, the strength of the relationship of the item with respect to the input evaluation term is less than or equal to a preset standard, and the category is The category having the strength of the relationship that is within the preset reference difference is changed as the next candidate. Since this next candidate is displayed instead of the initial candidate, the next candidate design having the same impression as the initial candidate but different atmosphere can be inferred in a short time.

According to the third aspect of the invention, when the design element of the next candidate is inferred and extracted, when the change to the next candidate includes a plurality of items, the category of the next candidate in the item is included. A combination between the items is extracted, and the change is performed for each combination. And
Since this next candidate is displayed instead of the initial candidate, the impression is the same as that of the initial candidate, and the next candidate design having a different atmosphere can be inferred in no time.

According to the fourth aspect of the invention, when the design element of the next candidate is inferred and extracted, when the change to the next candidate involves a plurality of items, the knowledge about the combination of categories for the evaluation terms is used. , Making the change to a category of matching combinations between the items.
Since this next candidate is displayed instead of the initial candidate, the next candidate design having the same impression as the initial candidate but different atmosphere can be inferred in a short time.

[Brief description of drawings]

FIG. 1 is a configuration diagram of the present invention.

FIG. 2 is an explanatory diagram showing adjective knowledge.

FIG. 3 is an explanatory diagram showing image knowledge.

FIG. 4 is an explanatory diagram showing image knowledge.

FIG. 5 is a basic flowchart.

FIG. 6 is a flowchart of next candidate inference according to the first embodiment.

FIG. 7 is an explanatory diagram showing a display example of a next candidate of the first embodiment.

FIG. 8 is an explanatory diagram showing combination knowledge of the second embodiment.

FIG. 9 is an explanatory diagram of next candidate inference according to the fourth embodiment.

FIG. 10 is an explanatory diagram of next candidate inference according to the fifth embodiment.

FIG. 11 is a configuration diagram of a conventional example.

[Explanation of symbols]

 1 keyboard (input means) 2 knowledge database 3 inference means 4th candidate inference means 5 CRT (display means) 9 evaluation terms 10 adjective knowledge 11 design elements 12 image knowledge 12a combination knowledge 16 computer

Claims (4)

[Claims] 1. A knowledge database that stores a plurality of items and a relationship between design elements classified into a plurality of categories for each item and design evaluation terms, input means for inputting the evaluation terms, and the input Inference means for inferring, for each item, a category in which the strength of relation to the evaluation term is maximum based on the evaluation term using the knowledge database, and display means for displaying the inferred design element in combination. In a design device comprising: a next candidate inference means for inferring and extracting a next candidate design element in place of the inferred design element, wherein the next candidate inference means selects a next candidate category for each item, Relationship between the inferred category and the next candidate category for each input item for each evaluation term Taking the difference in intensity, the item difference said strong of is within a preset reference difference, the design device and changes the categories mentioned above inferred next candidate categories. 2. The designing apparatus according to claim 1, wherein the next candidate inference means has a relationship strength of an item with respect to the input evaluation term that is equal to or less than a preset standard, and The design apparatus, wherein the category is changed for items whose difference in strength of the relationship is within a reference difference. 3. The designing apparatus according to claim 1, wherein the next candidate inference means selects a category including a category of the next candidate between the items when the change to the next candidate involves a plurality of items. A design device, wherein all combinations are extracted, and the change is made for each combination. 4. The design apparatus according to claim 1, wherein the next candidate inference means uses knowledge about a combination of categories for evaluation terms when a change to a next candidate involves a plurality of items. A design apparatus, characterized in that the category is selected and a change is made by selecting a suitable combination category.