JP6703412B2 - Aroma creation support method - Google Patents

Description

The present invention relates to a scent creation support method or a scent creation support system.

As for the scent of consumer products, it has been conventionally emphasized that a scent with a high preference is provided, and a selection method for selecting a scent with a high preference and a scent evaluation method for evaluating a scent with a high preference are proposed ( Patent Document 1).
In addition, there is proposed a method of classifying a term for evaluating an impression received from a scent into a sensory rating term and an emotion rating term to accurately evaluate a scent characteristic (Patent Document 2).

JP, 2007-63251, A JP, 2001-174450, A

It is known that a perfumer blends a scent by focusing on the character used to characterize the image by highlighting the image evoked by the scent. Generally, the mixed fragrance is composed of one or more characters, and the prescription of the mixed fragrance is created by adjusting the mixing ratio of the fragrance components belonging to each character.
Patent Literature 1 presents an evaluation method for the fragrance tone of the entire fragrance composition and a fragrance component that is a component contained in the fragrance composition, and does not present an evaluation method for characters.
Patent Document 2 uses a scent (lemon, rose, etc.) that evokes a unique character as a sample, but only evaluates each character individually, and what kind of character is selected and combined to obtain a blended fragrance. Or, there is no concrete presentation.

The present invention has been made in view of the above problems, and provides a fragrance creation support method or a fragrance creation support system used for creating a new blended fragrance by focusing on a character.

The present invention is a fragrance creation support method for creating a new mixed fragrance, wherein the existing mixed fragrance is added to the existing mixed fragrance based on an evaluation result indicating the extent to which the subject recalls the evaluation language. There is provided a perfume creation support method comprising a character extraction step of extracting one or more characters corresponding to at least one of an integrated language including a plurality of evaluation languages or one of the evaluation languages from a group of constituent characters. .

In addition, the present invention is a fragrance creation support system that supports the creation of a new fragrance preparation, wherein the existing fragrance preparation is based on an evaluation result indicating the degree to which the subject recalls the evaluation language for the existing fragrance preparation. A character extraction unit that extracts one or more characters corresponding to at least one of an integrated language including a plurality of the evaluation languages or one of the evaluation languages from a character group that constitutes a fragrance, and the character extraction unit. And an output unit that outputs one or more of the selected characters, and a fragrance creation support system.

According to the above invention, since a character suitable for recalling the evaluation language or the integrated language is extracted, the perfumer can easily create a blended fragrance that can recall a specific language based on the extracted character. You can

According to the present invention, there is provided a fragrance creation support method or a fragrance creation support system used for creating a new blended fragrance by focusing on a character.

It is a flow chart which shows a processing procedure of a fragrance creation support method concerning this embodiment. It is a figure which shows the model used for the covariance structure analysis in this embodiment. It is a figure which shows the combination of the character obtained from the result of the Y1 group which concerns on U1. It is a figure showing an example of the composition table of the character created based on the result of the Y1 group concerning U1. It is a figure showing an example of the composition table of the character created based on the result of the Y3 group concerning U1. It is a figure which shows the calculation result regarding a consumer survey. It is a figure which shows the calculation result regarding a consumer survey. It is a figure which shows briefly the structure of the production support system of a fragrance.

Embodiments of the present invention will be described below with reference to the drawings. Note that, in all the drawings, the same reference numerals are given to the same components, and the description thereof will be appropriately omitted.

<Regarding the processing procedure of the fragrance creation support method>
FIG. 1 is a flowchart showing a processing procedure of a fragrance creation support method according to the present embodiment.
The fragrance creation support method according to the present embodiment is for supporting the creation of a new mixed fragrance, and is determined based on the sensory evaluation of the existing mixed fragrance, and the evaluation result indicating the degree to which the subject recalls the evaluation language. From the integrated language including a plurality of evaluation languages or one or a plurality of characters corresponding to at least one of the evaluation languages is extracted from the character group constituting the existing blended fragrance. In the present embodiment, the process S5 shown in FIG. 1 corresponds to the character extraction process.

In the present invention, sensory evaluation is also referred to as sensory test or sensory test, and is to evaluate an object by using human senses (visual sense, hearing sense, taste sense, smell sense, touch sense). Is to evaluate the scent. The subject who performs the sensory evaluation may be a general consumer or a specialized panel. Further, the number of answers (the number of evaluation results) of the subjects required to carry out the present invention is not particularly limited.
The evaluation language is a language presented to the subject in the sensory evaluation, and in the present embodiment, an example in which 25 kinds of impression words and 16 kinds of emotion words, which will be described later, are used as the evaluation languages will be described. Note that this is a specific example, and the content and number of evaluation languages used for implementing the present invention are not particularly limited.
The integrated language is a language that expresses the meaning of including a plurality of evaluation languages, and is defined by an analyst who analyzes sensory evaluations in this embodiment. In particular, in the present embodiment, a language representing a meaning including a plurality of impression words may be referred to as an “impression factor”, but the “impression factor” is a specific example of an integrated language.

More specifically, the five steps (process S1 to process S5) shown in FIG. 1 will be described.

(Process S1)
The process S1 is a process of selecting a sample of the mixed flavor used in the product.
First, from the samples of the blended fragrances that have already been used in the commercial products, we selected the fragrance expressions to be diverse from the variations of the scents used in daily necessities. Examples of daily necessities include products used for the body such as shampoo, soap, antiperspirant spray, body lotion, and clothes and household products such as softening agents, laundry detergents, and space deodorants. From the variation of the scents used for these, the taste was not biased and it was selected from a wide range. Table 1 exemplifies 30 representative types of mixed fragrances. The scents used for evaluation in advance need not be 30 kinds.

(Processing S2)
The process S2 is a step of selecting an evaluation language to be used for the survey from the emotional words and the impression words that describe the scent. In the present embodiment, 16 kinds of emotion words shown in Table 2 and 25 kinds of impression words shown in Table 3 were selected as the evaluation languages.
Here, the impression word is a language related to the impression recalled by the subject in the sensory evaluation, and the emotional word is a language related to the emotion recalled by the subject in the sensory evaluation.

(Process S3)
In the process S3, the subject obtains the blended fragrance selected in the process S1 and the evaluation language selected in the process S2 from the image obtained by the subject from the scent of the target blended fragrance and the presented evaluation language (emotional word and impression word). Investigate the degree of agreement with the image.
When I smelled the selected mixed fragrance (Table 1), I gave a five-level evaluation as to whether or not to recall the scent image and emotion expressed in the evaluation language of Tables 2 and 3 (applicability). In addition to the items that I agree with, somewhat agree with, slightly agree with, and not agree with at all, I also asked the answer to select one item from these items.
As described above, the consumer survey in the process S3 is to obtain the evaluation result from the olfactory judgment of the general consumer who is the subject, which is one specific example of the sensory evaluation.

(Processing S4)
In step S4, an impression factor is extracted by covariance structure analysis based on the evaluation result obtained in step S3 (integrated language extraction step).
More specifically, in the process S4, a plurality of observation variables (hereinafter, referred to as a first observation variable group) representing an impression evaluation and a plurality of observation variables (hereinafter, referred to as a second observation variable group) representing an emotion evaluation. A model in which one or more latent variables are connected between and is analyzed by covariance structure analysis. Further, in process S4, when the model satisfies a predetermined criterion, the name given to the latent variable having a significant causal relationship with the observation variables forming the first observation variable group is used as an impression factor (integrated language). ).

In other words, the processing content of the processing S4 can be rephrased as the processing of analyzing the impression evaluation and the emotion evaluation and extracting the impression factor representing the specific impression having a high degree of contributing to the recall of the specific emotion as the integrated language. .. In addition, as described above, the impression factor is a language that represents a meaning including a plurality of impression words.
Here, the impression evaluation is an evaluation result that represents the degree of the impression that the subject remembers, which is determined by the sensory evaluation of the subject who sniffed the existing mixed fragrance, and the emotion evaluation is the degree of emotion that the subject recalls. Is an evaluation result representing. The impression evaluation in the example of the present embodiment is data obtained by digitizing the items selected in the 6-step evaluation of the impression word processed in the process S3, and the emotion evaluation is the emotion processed in the process S3. It is data that digitizes the items selected in the 6-level evaluation of words. For example, the impression word "refreshing scent" is given 5 points when "I think so much" is selected by the subject and 1 point when "I do not think at all" is selected by the subject. The average value of the number of points given by the process can be processed as an impression evaluation related to the impression word “fresh scent”.

Here, the covariance structure analysis is one of statistical analysis methods capable of examining the relationship between a plurality of constituent concepts.
In this study, when evaluating the image of human scent, the calculation was performed on the assumption of the cause of impression → emotion.

The process S4 will be described in detail below.
First, with respect to the evaluation result obtained in the process S3, in order to avoid the bias of the result of the covariance structure analysis performed next, first, a statistical process is performed to control the preference. Specifically, one-way ANOVA with different subjects was calculated by statistical analysis software SPSS (registered trademark). As a result of the calculation, three products with significantly different palatability were found, and therefore these were excluded from the analysis in the future.

Next, a covariance structure analysis was performed on the remaining 27 types of data. For the calculation, SPSS AMOS (trademark) which is a statistical analysis software was used. The validity of the calculation results was based on that the RMSEA value was below a predetermined threshold value and the GFI value was above a predetermined threshold value. The RMSEA value is a statistic that indicates poor relevance in the population. Also, it is said that the closer to 1.0 the GFI value is, the better the model.

FIG. 2 is a diagram showing a model used for the covariance structure analysis in this embodiment.
The rectangular object shown in FIG. 2 represents the observed variable. Here, the observation variable is actually measured data, and is either impression evaluation or emotion evaluation here. For example, the observation variable described as “adult scent” is the impression evaluation related to the impression word “adult scent”. Further, the observation variable described as "feeling like being happy" is an emotional evaluation related to the emotional word "feeling like being happy".
The elliptical object shown in FIG. 2 represents a latent variable. Here, the latent variable is a variable that has not been actually observed and is theoretically assumed. Note that U1 to U8 are shown as the names of latent variables in FIG. 2, but the names can be arbitrarily determined by the analyst, and like the observation variables, significant names may be given.
The object described as "eX (where X is a number)" shown in FIG. 2 represents an error variable. The magnitude relation of the numbers in X does not necessarily match the magnitude relation of the error variables.
The unidirectional arrow shown in FIG. 2 represents a causal relationship, and it is assumed that the variable at the origin of the arrow affects the variable at the end of the arrow. Although not shown in FIG. 2, path coefficients are given to all unidirectional arrows, and path coefficients given to unidirectional arrows from each latent variable to each observation coefficient are significant. It is assumed to be a value. That is, all the causal relationships from the latent variable shown in FIG. 2 to the observation coefficient are significant.
The double-headed arrow shown in FIG. 2 represents the correlation. Although not shown in FIG. 2, a correlation coefficient is given to each of the bidirectional arrows so that the magnitude of each correlation can be discriminated.

As shown in FIG. 2, there are four latent variables U1 to U4 that have a significant causal relationship with the observation variables forming the first observation variable group. In this covariance structure analysis, four U1 to U4 could be extracted as impression factors that evoke specific emotions.
Also, the impression factors U1 to U4 are for the observation variables related to the emotional words (second observation variable group) and the latent variables (U5 to U8) where a significant causal relationship is observed with the observation variables related to the emotional words. A causal relationship is directly or indirectly recognized. That is, the model shown in FIG. 2 clearly shows that the impression factors U1 to U4 contribute to arousing a specific emotion.

(Processing S5)
In process S5, the characters that form the blended fragrance are extracted based on the analysis result of the evaluation language performed in process S4. In order to perform character extraction, in advance, in order to analyze the elements constituting the blended fragrance selected in step S1, a scent character group is selected and sensory evaluation is performed on the blended fragrance using the scent character as an index. deep. Here, the type of the scent character is a scent character from the fragrance classification expression used by the perfumer, the expression of the perfume, and the typical scent material name with a characteristic scent character. Selected the most important ones. Table 4 exemplifies 50 types of representative scented character groups. The character group forming the scent need only be selected so that the scent expression can be varied, and need not be 50 types. With respect to all the scented character groups shown in Table 4, the odor intensity (sensory evaluation) was evaluated for all of the scented character groups shown in Table 4 for the mixed fragrance (Table 1) selected in the process S1.

Next, the existing mixed fragrances are classified into a plurality of groups based on the level of the evaluation value (score of the 6-level evaluation) of the subject related to the evaluation language given for each existing mixed fragrance.
More specifically, the observation variable data (impression result) included in each of the four impression factors obtained in the process S4 is given, and the score (factor score) of each impression factor for 27 kinds of mixed fragrances is given. I asked. In the present embodiment, the impression factor is made up of two observation variables, so the average value of these two points is used as the factor score. Then, 27 kinds of mixed fragrances were classified for each impression factor into a Y1 group having a high factor score, a Y2 group having an intermediate factor score, and a Y3 group having a low factor score. This is a process for facilitating extraction of a scent character characteristically associated with each impression factor in the rough set calculation described later.
Table 5 below shows the factor scores of each impression factor for 27 types of mixed fragrances. Further, Table 6 below shows the results of classifying 27 kinds of mixed flavors into three groups. In addition, “No.” shown in Tables 5 and 6 is an identification number given to each mixed flavor, and although it matches in Table 5 and Table 6, it is added to the sample of each mixed flavor in Table 1. "No." does not always match. Further, as shown in Table 6, the 27 kinds of compounded fragrances according to the present embodiment belong to any one group without duplication.

Next, in process S5, among rough character groups forming the existing mixed fragrance, rough set calculation is performed to treat the existing mixed fragrance forming one group out of the classified groups (Y1 group to Y3 group) as the same grade. Extract one or more characters from.
More specifically, rough set calculation was performed for the three classified groups. For the calculation, a dedicated calculation software (“Ron set analysis program with grade” by Holon Create Co., Ltd.) was used.
This software has a limit to the number of data sets that can be calculated, and the variables are up to 17. Therefore, among the 50 types of characters that make up the mixed fragrances included in the Y1 group, 17 types were divided into three, such as a top note, 17 types of middle notes, and 16 types of base notes for calculation. In normal rough set calculation, numerical values are treated as categorical data, but in graded rough set calculation, numerical weighting can be reflected in the calculation to make ordinal data. Therefore, the odor intensity of each of the six levels obtained in the process S1 is set as 0→0, 1→0.2, 2→0.4, 3→0.6, 4→0.8, 5→1. Converted

The rough set is one of the analytical methods proposed by Pawlak. Z in 1982, and is currently applied in many experiments of Kansei engineering. The characteristic of the analysis is "to capture the set of objects well (roughly) and grasp the features."
If this calculation method is used, it is required that each character listed in the scent profile contributes to the corresponding specific impression. Compared with the conventional consideration in the multiple regression analysis, if the formula U1=a1 (citrus) X1+a2 (Woody) X2+b (a1>0 and a2>0) is derived, increasing the amount of citrus results in U1 The impression concerning can be increased. That is, in such an expression, the weighting of the main character can be discussed, but it is difficult to obtain the elements other than the main character and it is difficult to discuss the balance. In addition, due to the nature of olfaction, it is not possible to expect a linear increase in the value of U1 even if the number of elements is increased (saturation of olfaction occurs). On the other hand, rough set theory is effective for qualitative data, and the calculation results can be discussed in a combination such as (citrus) + (woody). In addition, it is also an advantage of using rough sets that perfume can be expected to develop by using the accord rather than using one material. It is expected that such calculation results will be useful for communication with perfumer and knowledge of fragrance development. In addition, the evolutionary theory of graded rough sets can introduce an interval measure for the quantity of elements, which complements both the extraction of elements and the approximate quantity.

FIG. 3 shows a part of the result of rough set calculation in this embodiment. FIG. 3 is a diagram showing a combination of characters obtained from the result of the Y1 group related to U1. That is, FIG. 3 is a diagram showing a combination of characters that enhance the impression of U1.
As a result of the calculation, for each impression factor, a combination of characters belonging to each of a top note (denoted by T in FIG. 3), a middle note (denoted by M in FIG. 3), and a base note (denoted by B in FIG. 3) was gotten. Regarding the calculation results having common items, the calculation results are further summarized. For example, in the Y1 group related to U1, a plurality of elements such as [>4 Minty] and [>3 Lemon, >2 Orange] are extracted.
Here, the top note is a character that is highly volatile and easy to fly, and plays a role of giving the fragrance a characteristic such as refreshing the mouth of the scent. Characters corresponding to the top notes include, for example, citrus, green, fruity, and the like.
Here, the middle note is a character with moderate volatility and retention, and is often the main feature of the fragrance. Characters corresponding to middle notes include, for example, rose, jasmine, muguet, and the like.
Here, the base note is a character with low volatility and high retentivity, and plays a role of remaining (remaining scent). Characters corresponding to the base note include, for example, Woody, Mussky, and the like.

As shown in FIG. 3, it was found from the result of the Y1 group related to each impression factor (U1 to U4) that a combination of a plurality of scent characters that can easily contribute to the improvement of each impression factor was obtained. In addition, it was found from the results of the Y3 group related to each impression factor (U1 to U4) that a combination of a plurality of scent characters that is hard to contribute to the improvement of each impression factor is obtained. Based on this result, a character composition table was created.
FIG. 4 is a diagram showing a specific example of a character configuration table created based on the result of the Y1 group related to U1. That is, FIG. 4 is a diagram showing a configuration table of a character that is a source of the mixed fragrance for enhancing the impression of U1. FIG. 5 is a diagram showing a specific example of a character configuration table created based on the result of the Y3 group related to U1. That is, FIG. 5 is a diagram showing a configuration table of a character that is a source of the mixed fragrance that does not enhance the impression of U1.
In the configuration table of FIG. 4 or FIG. 5, the strength of the character is indicated by the number of shaded columns among the columns indicating one or more. In addition, a shaded column in the column indicating 0 indicates that a strength less than 1 is allowed for the character. In addition, a black-painted column in the column indicating 0 indicates that the character is not included in the configuration.

<Verification of blended fragrances based on character composition table>
By using the composition table of the character obtained in the process S5 described above, the perfumer can efficiently create the mixed perfume (perfume composition) that effectively enhances the specific impression.
In order to verify this, for each of the impression factors U1 to U4, the perfumer selects the fragrance material corresponding to each character, creates the base of the fragrance, and modifies or supplements the fragrance to adjust the base scent. Was added to complete the compounded perfume for the product. Then, a consumer survey was conducted using the prepared fragrance for the product. It was decided that the scene for sniffing the scent was "a scene for sniffing a body care product" (a body care product is an out-bath product that does not wash off, and it is clearly indicated that body lotion, body cream, body milk, etc. are included). The subjects were 106 Japanese women in their 20s to 50s who performed body care more than twice a week within 3 months. As the quality survey product, a non-breathable sheet (4 cm image) carrying perfume oil was enclosed in an aluminum pillow, and the test subject was told to "open the aluminum pillow firmly and keep the nose about 5 cm away from the aluminum pillow." Evaluate the scent at the distance and at the time and place you normally use body care products." The degree of applicability of the question item is rated on a scale of 5, and very strongly, yes, slightly yes, slightly yes, not at all, and in addition to the question item, 1 out of 1 I asked for an answer to select an item.

FIG. 6 and FIG. 7 are diagrams showing calculation results regarding the above consumer survey, in which the left side of each bar graph is derived from Y1 and the right side is derived from Y3. Specifically, FIG. 6(a) compares the results of a consumer survey on the compounded fragrance prepared based on the result of the Y1 group related to U1 and the compounded fragrance created based on the result of the Y3 group related to U1. FIG. FIG.6(b) is a figure which compares the result of the consumer survey regarding the blended fragrance|flavor created based on the result of the Y3 group which concerns on U2, and the blended fragrance|flavor created based on the result of Y1 group which concerns on U2. is there. FIG. 7( a) is a diagram comparing the results of consumer surveys regarding the blended fragrance created based on the result of the Y1 group related to U3 and the blended fragrance created based on the result of the Y3 group related to U3. is there. FIG.6(b) is a figure which compares the result of a consumer survey regarding the compounded fragrance|flavor created based on the result of the Y3 group concerning U4, and the compounded fragrance|flavor created based on the result of the group Y3 related to U4. is there.
In addition, the vertical axis in the figure is 5 points for "I think so", 4 points for "I think so", 3 points for "I think so", and "Slightly think so" The average value of each item is shown in the case of giving 2 points for the answer of "I do not think so", 1 point for the answer of "I do not think so", and 0 point for the answer of "I do not know". The bar in the figure represents the standard deviation.
As a result of the consumer survey, when the prepared fragrance is prepared based on U1, U2 and U3 among the prepared fragrances, the prepared fragrance prepared by using the character of Y1 group uses the character of Y3 group. The results were significantly higher than those of the prepared fragrances (ρ<0.05). On the other hand, regarding the impression factor of U4, no significant difference was found between the two samples.

As described above, the present inventor diligently studied a method for more efficiently designing a strong impression on consumers. As a result, even among the scents with a high preference evaluation, when the characters were classified based on the impression factor associated with a particular impression word, the characters with high contribution to each impression factor were found.
This makes it possible to design a scent that gives a strong impression more efficiently by using those characters.

It is possible to construct a fragrance creation support system that supports the creation of a new mixed fragrance by utilizing the above fragrance creation support method. FIG. 8 is a diagram schematically showing the configuration of the fragrance creation support system 10.
The fragrance creation support system 10 includes a character extraction unit 11, a storage unit 12, a network 13, an information processing terminal 14, and a printing device 15.
The character extraction unit 11 selects an integrated language that includes a plurality of evaluation languages from the character group of the existing compounded flavor, based on the evaluation result indicating the degree to which the subject recalls the evaluation language for the existing compounded flavor. One or more characters corresponding to at least one of the evaluation languages are extracted.
The storage unit 12 may store at least one or more characters extracted by the character extraction unit 11 and may store other data processed by the character extraction unit 11. In addition, the character stored in the storage unit 12 is updated so that the evaluation results input to the character extraction unit 11 are accumulated, and the character has an accuracy enough to recall the corresponding evaluation language or integrated language. Is preferably improved.
The network 13 connects the storage unit 12 and an external device (for example, the information processing terminal 14 or the printing device 15) in a communicable manner, and plays a role of allowing the data stored in the storage unit 12 to be spread over a wide range. Fulfill. The network 13 is, specifically, the Internet, a LAN (Local Area Network), an information communication line such as a mobile phone communication network, and hardware for constructing the line.
The information processing terminal 14 can acquire the data (including the extraction result by the character extraction unit 11) stored in the storage unit 12 via the network 13 and display and output the acquired data. Further, the printing device 15 can acquire the data stored in the storage unit 12 via the network 13 and print out the acquired data. That is, the information processing terminal 14 and the printing device 15 can function as an output unit that outputs one or a plurality of characters extracted by the character extraction unit 11. The mode of output by the information processing terminal 14 or the printing device 15 is not particularly limited, and for example, a list showing combinations of characters as shown in FIG. 3 may be output, or the list shown in FIG. 4 or FIG. It is also possible to output a composition table of a character that enhances or does not enhance the impression according to a specific impression factor as shown.

<Modification>
The present invention is not limited to the above embodiments, and the contents of the above embodiments may be modified or improved as long as the object of the present invention is achieved.
For example, an existing evaluation result may be used instead of the evaluation result obtained by the process S1, the process S2, and the process S3, and may not necessarily be included in the process of the fragrance creation support method.
Further, in the above-described embodiment, the number of subjects is set, but, for example, the present invention is implemented by performing the same processing on the evaluation result based on the sensory evaluation of one famous perfumer. Good.

Although the processing S4 (integrated language extraction step) according to the above-described embodiment has been described as an example in which the integrated language is extracted by the covariance structure analysis, the analysis method of the integrated language extraction step is not limited to this example. For example, the integrated language can be derived by processing the evaluation result obtained by the sensory evaluation by using multivariate analysis.
Further, although the processing S5 (character extraction step) according to the above-described embodiment has described the example in which the combination of characters is extracted by the rough set calculation, the analysis method of the character extraction step is not limited to this example. For example, the character may be derived by processing using multiple regression analysis.

In the process S5 (character extraction process) according to the above-described embodiment, one or more characters corresponding to the impression factors (U1 to U4) that are the integrated language extracted in the process S4 (integrated language extraction process) are added to the existing characters. The example of extracting from the character group that constitutes the blended fragrance has been described. However, this embodiment is a specific example, and the practice of the present invention is not limited to this.
For example, the character extracted in the process S5 may include one or a plurality of characters corresponding to only one evaluation language (impression word or emotional word) selected in the process S2. At this time, if the one or more characters extracted in the process S5 correspond to the one evaluation language selected in the process S2, the process S4 (integrated language extraction process) is not necessarily the process of the fragrance creation support method. Need not be included in.

In the above-described embodiment, an example in which an impression word or an emotional word is used as an evaluation language has been described, but a language similar thereto, that is, a wide range of languages relating to human emotions and sensibilities may be treated as an evaluation language.

In the above-described embodiment, the present invention has been described as a method for supporting the creation of a fragrance, but if the process up to the process of blending a fragrance based on the character (character composition table) extracted in step S5 is included in a series of steps, The above-described embodiment can also be treated as a method for creating a fragrance.

Only the feature of the processing S4 in the above-described embodiment can be regarded as a unique invention. That is, in the process S4, the impression evaluation and the emotion evaluation are analyzed, and the impression factor (integrated language) representing the particular impression having a high degree of contributing to the recall of the particular emotion is extracted. Based on this, it is possible to assist the design of the base composition of the blended fragrance that enhances a specific impression.

The fragrance creation support system 10 described in the above embodiment has been described as a communication network system constructed via the network 13, but this configuration is one specific example, and the implementation of the present invention is not limited to this. . For example, a simple configuration in which the character extraction unit 11 inputs the evaluation result in the sensory evaluation and outputs one or a plurality of characters that are the extraction result may be adopted.

This embodiment includes the following technical ideas.
(1) Based on the evaluation result indicating the degree to which the subject recalls the evaluation language for the existing mixed flavor, which is determined based on the sensory evaluation of the existing mixed flavor, among the characters constituting the existing mixed flavor A method for assisting creation of perfume, comprising: a character extraction step of extracting one or a plurality of characters corresponding to at least one of an integrated language including a plurality of the evaluation languages or one of the evaluation languages.
(2) An integrated language extraction step of analyzing the evaluation result and extracting the integrated language is provided, and in the character extraction step, one or a plurality of characters corresponding to the integrated language extracted in the integrated language extraction step are extracted. The method of supporting creation of a fragrance according to (1), wherein the character is extracted from a group of characters constituting the existing mixed fragrance.
(3) The evaluation language includes an impression word relating to an impression recalled by the subject and an emotion word relating to an emotion recalled by the subject, and is determined by the sensory evaluation in the integrated language extraction step. By analyzing an impression evaluation indicating the degree to which the subject recalls the impression word and an emotion evaluation indicating the degree to which the emotion word is recalled, a specific impression that is highly contributed to the recall of a specific emotion is expressed, and The method for supporting creation of a fragrance according to (2), wherein the integrated language including a plurality of the impression words is extracted.
(4) In the integrated language extraction step, one or a plurality of observation variable groups, which are a plurality of observation variables representing the impression evaluation, and a second observation variable group, which are a plurality of observation variables representing the emotion evaluation, are connected to each other. A model connected by a plurality of latent variables is analyzed by a covariance structure analysis, and when the model satisfies a predetermined criterion, a significant causal relationship is observed with respect to the observation variables constituting the first observation variable group. The fragrance creation support method according to (3), in which the name given to the latent variable is extracted as the integrated language.
(5) In the character extraction step, the existing compounded fragrances are classified into a plurality of groups based on the level of the evaluation value of the subject related to the evaluation language given to each of the existing compounded fragrances, and the group is classified. One or a plurality of characters are extracted from the character group forming the existing mixed fragrance by rough set calculation which treats the existing mixed fragrance forming one of the groups as the same grade (from (1) ( The method for supporting the preparation of a fragrance according to any one of 4).
(6) A fragrance creation support method for creating a new mixed fragrance, which is an impression evaluation and emotional expression indicating a degree of impression evoked by a subject determined by a sensory evaluation of a subject who sniffed an existing mixed fragrance. A method for supporting the creation of a fragrance, comprising a step of analyzing an emotion evaluation indicating a degree and extracting an integrated language expressing a specific impression having a high degree of contributing to recalling a specific emotion.
(7) A fragrance creation support system for creating a new fragrance composition, wherein the existing fragrance composition is configured based on an evaluation result indicating the degree to which a subject recalls the evaluation language for the existing fragrance mix. A character extraction unit for extracting one or more characters corresponding to at least one of an integrated language including a plurality of the evaluation languages or one of the evaluation languages from the character group, and one extracted by the character extraction unit. Alternatively, a fragrance creation support system including an output unit that outputs a plurality of characters.

10 Fragrance Creation Support System 11 Character Extractor 12 Storage 13 Network 14 Information Processing Terminal 15 Printing Device

Claims (1)

Based on a sensory evaluation using an existing compounded fragrance, a fragrance creation support method that supports the creation of a new compounded fragrance by extracting one or more scent characteristics from a predetermined fragrance feature group. There
The existing compounded fragrance is one in which the odor intensity is evaluated in advance for each of the scent characteristic groups,
The sensory evaluation, for a subject who sniffed the existing compounded fragrance, presents a plurality of predetermined evaluation languages, and one of a plurality of levels regarding the degree of agreement between the scented compounded fragrance and the presented evaluation language. Depending on the format to be selected, an answer was obtained from the subject for each of the plurality of evaluation languages,
In the plurality of evaluation languages presented to the subject in the sensory evaluation, a plurality of impression words with a scent added to the end of the expression expressing an impression that can be recalled from the scented mixed fragrance, and the subject from the scented mixed fragrance Includes a plurality of emotion words that represent emotions to recall,
An analyst performs a covariance structural analysis using a computer based on the observation variables defined for each evaluation language by digitizing the stage selected by the subject for the evaluation language used for the sensory evaluation, at least A significant causal relationship is directly recognized with respect to the observation variables related to the two impression words, and significant directly or indirectly with respect to the observation variables related to the emotion words via another latent variable. A latent variable having a causal relationship is extracted, and one extracted latent variable is set as an impression factor including a plurality of the impression words related to the observed variable having a directly significant causal relationship with the latent variable. The first step,
The computer scores a factor score for each of the existing blended fragrances based on the observed variable that is directly and significantly associated with the latent variable that is the impression factor in the first step. By calculating a factor score for each of the existing compounded fragrances that have been calculated and a calculation process based on the odor intensity evaluated for each of the existing compounded fragrances. A second step of extracting a combination of the scent characteristics that enhances each of the impressions represented by the impression words,
A method for supporting the creation of fragrances.