WO2024071072A1 - Material recommendation device, material recommendation method, and program - Google Patents

Abstract

This material recommendation device comprises: an input unit that receives the input of information indicating a product type and emotional language that indicates language representing an emotion; a material parameter extraction unit that extracts the value of a sensation parameter for each material corresponding to the product type on the basis of material parameter information in which the material used in a product belonging to the product type and the value of the sensation parameter, which indicates the degree of sensation a human receives with respect to the material, are associated with each other; an emotional language parameter extraction unit that extracts the value of the sensation parameter corresponding to the emotional language on the basis of emotional language parameter information in which the emotional language and the value of the sensation parameter are associated with each other; and a recommended material determination unit that determines a recommended material for the emotional language on the basis of the value of the sensation parameter for each material extracted by the material parameter extraction unit and the value of the sensation parameter extracted by the emotional language parameter extraction unit.

Description

Material recommendation device, material recommendation method and program

The present invention relates to a material recommendation device, a material recommendation method, and a program.
This application claims priority based on Japanese Patent Application No. 2022-152385, filed on September 26, 2022, the contents of which are incorporated herein by reference.

Indicators that users use as evaluation criteria when selecting a product include, for example, the characteristics and performance of the materials used in the product. In addition, the impression that users get, such as how the material is perceived through the five senses (sight, touch, hearing, taste, smell) and how the material appeals to the senses, are also important indicators that users use as evaluation criteria when selecting a product.

　Previously, it was difficult to accurately present materials that would produce the sensations desired by the user. In order to recommend materials that match the sensations desired by the user, for example, the person in charge of recommending the materials needs to thoroughly interview the user. In such cases, however, whether or not materials that produce the sensations desired by the user are appropriately recommended also depends heavily on the sense of the person in charge. For example, even if the user is looking for "comfortable" materials, what the user considers comfortable does not necessarily match what the person in charge considers comfortable.

To address this issue, for example, the similar texture material recommendation system described in Patent Document 1 stores parameter values in advance that indicate the correspondence between terms (onomatopoeia) that express texture and the physical properties of the surface of a material, and automatically presents recommended materials for an input onomatopoeia based on the parameter values.

Patent No. 4469908

The similar tactile material recommendation system described in Patent Document 1 is a system that only accepts input using onomatopoeia that expresses the sense of touch. In reality, however, the impression a user gets from a material is not necessarily an impression obtained through the sense of touch. The impression a user gets from a material may include impressions obtained through multiple or all of the five senses.

In addition, when users express the impression they get from an ingredient in words, they do not necessarily use onomatopoeias (such as "smooth" or "slippery") that simply mimic the impression they get from the five senses. Users often use words that express emotions to describe the impression they get from an ingredient. Words that express emotions here are adjectives that are used to quantify human sensations in general sensory evaluations. Some of these adjectives are not limited to a specific sensation among the five senses, such as "favorite," "unique," "comfortable," and "luxurious." For these reasons, conventional technology has had the problem of making it difficult to recommend ingredients based on people's emotions.

The present invention was made in consideration of these circumstances, and aims to provide an ingredient recommendation device, ingredient recommendation method, and program that can recommend ingredients according to a person's emotions.

The subject matter of the present disclosure includes the following embodiments.
[1] An ingredient recommendation device comprising: an input unit that accepts input of information indicating a product type and an emotion word indicating a word expressing an emotion; an ingredient parameter extraction unit that extracts a sensory parameter value for each ingredient corresponding to the product type input to the input unit based on ingredient parameter information in which materials used in products belonging to the product type are associated with sensory parameter values indicating the degree of sensation a human feels for the material; an emotion word parameter extraction unit that extracts a sensory parameter value corresponding to the emotion word input to the input unit based on emotion word parameter information in which the emotion word is associated with the sensory parameter values; and a recommended ingredient determination unit that determines the ingredient recommended for the emotion word input to the input unit based on the sensory parameter value for each ingredient extracted by the ingredient parameter extraction unit and the sensory parameter value extracted by the emotion word parameter extraction unit.

[2] The material recommendation device described in [1], wherein the emotion words are words classified as evaluation factors or activity factors among factors of the EPA (Evaluation, Potency, Activity) structure defined in impression evaluation by the SD (Semantic Differential) method.

[3] The material recommendation device according to [1] or [2], further comprising a correction unit that corrects the values of the sensory parameters using coefficients that take into account the cross-modal effect in which impressions received through multiple types of senses are related to each other.

[4] The material recommendation device described in [3], in which the coefficients are set to different values depending on the order in which impressions are received by the multiple types of senses, and the correction unit acquires information indicating the order and performs correction using the coefficients corresponding to the acquired order.

[5] The material recommendation device described in [4], in which the coefficient used to correct the value of the sensory parameter for each material is set to a value of 1 for the sensation received first, a value greater than 0 and less than 1 for the sensation received later, and a value of 0 for any other sensation.

[6] The material recommendation device according to [4] or [5], in which the coefficients used to correct the values of the sensory parameters of the emotional words are set to a value of 1 for the senses in which the cross-modal effect occurs, and are set to a value of 0 for the senses in which the cross-modal effect does not occur.

[7] The material recommendation device according to any one of [1] to [6], wherein the senses are the five senses or a further subdivided sense of the five senses.

[8] A material recommendation method comprising: an input step of accepting input of information indicating a product type and an emotion word indicating a word expressing an emotion; a material parameter extraction step of extracting a sensory parameter value for each material corresponding to the product type input in the input step based on material parameter information in which materials used in products belonging to the product type are associated with sensory parameter values indicating the degree of sensation a human feels for the material; an emotion word parameter extraction step of extracting a sensory parameter value corresponding to the emotion word input in the input step based on emotion word parameter information in which the emotion word is associated with the sensory parameter value; and a recommended material determination step of determining the material recommended for the emotion word input in the input step based on the sensory parameter value for each material extracted in the material parameter extraction step and the sensory parameter value extracted in the emotion word parameter extraction step.

[9] A program for causing a computer to execute the following steps: an input step of accepting input of information indicating a product type and an emotion word indicating a word expressing an emotion; a material parameter extraction step of extracting a sensory parameter value for each material corresponding to the product type input in the input step based on material parameter information in which materials used in products belonging to the product type are associated with sensory parameter values indicating the degree of sensation a human feels for the material; an emotion word parameter extraction step of extracting a sensory parameter value corresponding to the emotion word input in the input step based on emotion word parameter information in which the emotion word is associated with the sensory parameter value; and a recommended material determination step of determining the material recommended for the emotion word input in the input step based on the sensory parameter value for each material extracted in the material parameter extraction step and the sensory parameter value extracted in the emotion word parameter extraction step.

The present invention makes it possible to recommend ingredients based on a person's emotions.

1 is a block diagram showing a functional configuration of a material recommendation device 100 according to an embodiment of the present disclosure. FIG. 2 is a diagram showing an example of an emotion word input to the material recommendation device 100 according to an embodiment of the present disclosure. FIG. 2 is a diagram illustrating an example of five sense parameters used in the material recommendation device 100 according to an embodiment of the present disclosure. 1 is a diagram showing an example of material parameter information 111 stored in a parameter information storage unit 110 of a material recommendation device 100 according to an embodiment of the present disclosure. FIG. FIG. 2 is a diagram showing an example of emotion word parameter information 112 stored in a parameter information storage unit 110 of the material recommendation device 100 according to an embodiment of the present disclosure. FIG. 2 is a diagram illustrating an example of modal correction information 113 stored in a parameter information storage unit 110 of the material recommendation device 100 according to an embodiment of the present disclosure. FIG. 2 is a diagram illustrating an example of modal correction information 113 stored in a parameter information storage unit 110 of the material recommendation device 100 according to an embodiment of the present disclosure. 4 is a flowchart showing an operation of the material recommendation device 100 according to an embodiment of the present disclosure. 11A and 11B are diagrams illustrating an example of a correction process for material parameter values performed by the modal correction unit 133 of the material recommendation device 100 according to an embodiment of the present disclosure. FIG. 11 is a diagram illustrating an example of a correction process for the values of emotion word parameters performed by a modal correction unit 133 of the material recommendation device 100 according to an embodiment of the present disclosure. FIG. 11 is a diagram illustrating an example of a calculation process of a distance between coordinates performed by a recommended material determination unit 134 of the material recommendation device 100 according to an embodiment of the present disclosure. 1 is a diagram showing an example of information output by an output unit 140 of the material recommendation device 100 according to an embodiment of the present disclosure. FIG. FIG. 13 is a diagram showing another example of information output by the output unit 140 of the material recommendation device 100 according to an embodiment of the present disclosure.

Below, one embodiment of the present disclosure will be described with reference to the drawings.

[Configuration of material recommendation device]
The following describes the configuration of the material recommendation device 100 according to the embodiment. Fig. 1 is a block diagram showing the functional configuration of the material recommendation device 100 according to an embodiment of the present disclosure.

The material recommendation device 100 is an information processing device configured using, for example, a general-purpose computer. The material recommendation device 100 accepts input of words expressing the user's desired feeling for the materials used in the product and expressing the user's emotions (hereinafter referred to as "emotion words"). The material recommendation device 100 determines at least one recommended material based on the input emotion words, and outputs information indicating the determined material.

FIG. 1 is a block diagram showing the functional configuration of a material recommendation device 100 in a first embodiment of the present disclosure. As shown in FIG. 1, the material recommendation device 100 includes a parameter information storage unit 110, an input unit 120, a material parameter extraction unit 131, an emotion word parameter extraction unit 132, a modal correction unit 133, a recommended material determination unit 134, and an output unit 140.

The parameter information storage unit 110 is configured using a computer-readable storage medium. The computer-readable storage medium referred to here is, for example, a magnetic hard disk device or a semiconductor storage device. The parameter information storage unit 110 pre-stores various types of parameter information used to determine recommended materials. As shown in FIG. 1, the parameter information storage unit 110 stores at least material parameter information 111, emotion word parameter information 112, and modal correction information 113. Details of each piece of parameter information will be explained later.

The input unit 120 is configured using an input device that accepts input operations by a user. The input device here is, for example, a keyboard, a mouse, a touch panel, and an input button. As shown in FIG. 1, the input unit 120 is configured to include a product type input unit 121, an emotion word input unit 122, and a usage status input unit 123.

The product type input unit 121 acquires information indicating the product type input by the user (hereinafter referred to as "product type information"). The product type input unit 121 outputs the acquired product type information to the material parameter extraction unit 131. The product type information here is information indicating the type of material used in the product, such as, for example, "food packaging material," "housing building material," and "synthetic leather material for bags."

The emotion word input unit 122 acquires information indicating emotion words input by the user (hereinafter simply referred to as "emotion words"). The emotion word input unit 122 outputs the acquired emotion words to the emotion word parameter extraction unit 132. The emotion words referred to here are words such as adjectives used to quantify human sensations in general sensory evaluations. Such adjectives include words that are not limited to a specific sensation among the five senses, such as "favorite," "unique," "comfortable," and "luxurious."

In general sensory evaluation, when human emotions are quantified and evaluated, the Semantic Differential (SD) method is often used. The SD method is a technique for measuring a person's impression (sensitivity) of an object using paired adjectives with opposite meanings, and evaluating it numerically. The SD method is characterized by its ability to express the structure of an impression with a small number of factors. For example, the SD method finds a structure consisting of three basic factors that can be interpreted as an evaluation factor, an activity factor, and a potency factor. This structure is called the EPA (Evaluation, Potency, Activity) structure.

Evaluation factors are expressed by pairs of adjectives that express a comprehensive evaluation of the value of an object, such as "like-dislike," "good-bad," and "beautiful-ugly." In addition, evaluation factors are also composed of pairs of adjectives that express the nature of an object, such as "stable-unstable," "clear-muddy," and "warm-cold," although this differs depending on the type of object being evaluated (shape, color, sound, etc.).

Both activity and potency factors are composed of expressions about the properties of an object. Activity factors are expressed by adjective pairs such as "noisy-quiet," "dynamic-static," and "flashy-subdued." Potency factors are expressed by adjective pairs such as "hard-soft," "sharp-dull," and "tense-relaxed."

FIG. 2 is a diagram showing an example of emotion words input to the material recommendation device 100 in one embodiment of the present disclosure. Of the three basic factors, the evaluative factor and the active factor contain many adjectives that directly express human emotions. In this embodiment, as an example, among the factors of the EPA structure defined in impression evaluation using the SD method, words classified as the evaluative factor and the active factor are defined as emotion words. Note that rather than defining pairs of adjectives as emotion words, both pairs of adjectives are defined separately as one emotion word.

The usage status input unit 123 acquires information indicating the usage status of the product (hereinafter referred to as "usage status information") input by the user. The usage status input unit 123 outputs the acquired usage status information to the modal correction unit 133.

The usage information referred to here is information that can identify which of the five senses the user receives when receiving an impression from a product. Furthermore, when a user receives impressions through more than one of the five senses, the usage information is information that can identify the order in which the impressions are received through each sense. For example, when a user receives impressions through two of the five senses, sight and touch, the usage information is information that can identify whether the user sees the product first and then touches it, or whether the user touches the product first and then sees it.

The material parameter extraction unit 131, the emotion word parameter extraction unit 132, the modal correction unit 133, and the recommended material determination unit 134 are functional units that are realized by a processor such as a CPU (Central Processing Unit) reading and executing a program that is pre-stored in a storage unit (not shown), for example. Note that the above storage unit and the parameter information storage unit 110 may be configured using the same storage medium.

The material parameter extraction unit 131 acquires the product type information output from the product type input unit 121. The material parameter extraction unit 131 also refers to the material parameter information 111 stored in the parameter information storage unit 110. The material parameter extraction unit 131 extracts, from the material parameter information 111, the values of the material parameters associated with the product type that corresponds to the acquired product type information.

The material parameter information 111 is information that indicates the degree of association between a material and the five sense parameters. The five sense parameters referred to here are composed of items that indicate the five senses, or items that indicate sensations that are further subdivided from each of the five senses. In this embodiment, the five sense parameters are composed of items that indicate sensations that are further subdivided from each of the five senses.

FIG. 3 is a diagram showing an example of the five sense parameters used in the material recommendation device 100 in one embodiment of the present disclosure. As shown in FIG. 3, touch is further classified into the five sense parameters of "hardness," "roughness," "friction," and "warm/cold." Vision is further classified into the five sense parameters of "color 1," "color 2," "color 3," "texture 1," "texture 2," and "texture 3."

It is known that touch can be classified into four elements: hardness, roughness, friction, and warmth/coldness. More specifically, for example, the values of average roughness Ra and maximum height Rz defined by the standard (JIS B 0601) may be used as the five sensory parameters for touch "roughness". Alternatively, parameters and values obtained by subjecting the average roughness Ra and maximum height Rz, etc. to principal component analysis or factor analysis may be used to reduce the dimensions. For example, the value of Young's modulus calculated based on the pressing force and displacement may be used as the five sensory parameters for touch "hardness". For example, the value of the dynamic friction coefficient when touching a material may be used as the five sensory parameters for touch "friction".

When the material recommendation device 100 is adapted to various materials, it is desirable to obtain uniformly evaluated parameter values for all materials with respect to touch, for example using a wearable device. By attaching a wearable device to a person's hand and actually touching the material, friction, hardness, etc. can be uniformly measured.

With regard to vision, the five sense parameters of vision shown in FIG. 3 include "Color 1", "Color 2", and "Color 3", but it is generally known that color can be expressed using three variables, such as L*a*b* or L*C*H. For example, these three variables may be used as the five sense parameters of "Color 1", "Color 2", and "Color 3", respectively. Note that the material recommendation device 100 may use the three variables in any color space, but it is necessary to evaluate all materials using the three variables in the same color space.

It is not known that vision can be clearly classified into a finite number of elements, as is the case with touch. However, it has recently become clear that color and texture elements are important factors that affect people's visual impressions. Here, texture refers to physical property information related to reflectance, such as the bidirectional reflectance distribution function (BRDF). BRDF (Bidirectional Reflectance Distribution Function) is the angular distribution characteristic of reflected light when light is incident from a specific angle. Generally, in CG (Computer Graphics), the texture of a material is reconstructed by rendering using the parameters of this BRDF. Of course, with regard to these parameters, parameters and values that have been reduced in dimension by principal component analysis or factor analysis may also be used.

With regard to taste, the type and amount of taste chemical components are thought to be factors that affect people's impressions. With regard to smell, the type and concentration of olfactory chemical components are thought to be factors that affect people's impressions. With regard to hearing, the signal strength and power spectrum are thought to be factors that affect people's impressions. Of course, with regard to these parameters as well, parameters and values that have been reduced in dimension by principal component analysis or factor analysis may be used.

The value of the material parameter may be, for example, the actual measurement value obtained by some measurement method, or the material parameter value of all materials may be a standardized value. However, if the actual measurement value is used, the unit of the material parameter value for all materials must be unified. For example, if the average roughness Ra value specified by the standard is used as the material parameter of "roughness", the material parameter value of "roughness" for all materials should be unified to a unit value such as millimeters or micrometers.

In addition, among the five senses, the impressions made by the materials used in industrial products such as food packaging materials, building materials for homes, and synthetic leather for bags are thought to be weighted heavily by the senses of sight and touch.

FIG. 4 is a diagram showing an example of material parameter information 111 stored in the parameter information storage unit 110 of the material recommendation device 100 in one embodiment of the present disclosure. The material parameter information 111 is data indicating the association between a material and the five sensory parameters described above. Hereinafter, the five sensory parameters associated with a material will be referred to as "material parameters."

In the material parameter information 111 shown in FIG. 4, the product type, product number, information indicating the material composition of the product, and material parameters are associated with each other. Each row of the material parameter information 111 is information about one material.

For example, in the first line of the material parameter information 111 shown in FIG. 4, the value "food packaging material" is registered in the "product type" field, and the value "101a001b001" is set in the "product number" field. This indicates that the material to which the product number "101a001b001" is assigned is a material used for food packaging material. Also, in the first line of the material parameter information 111 shown in FIG. 4, the value "101" is set in the "main component" field of the "product material composition" field, the value "a001" is set in the "additive 1" field, and the value "b001" is set in the "additive 2" field. This indicates that the material to which the product number "101a001b001" is assigned is a material that includes a main component assigned with the identification information "101", an additive assigned with the identification information "a001", and an additive assigned with the identification information "b001".

Furthermore, for each material parameter in the material parameter information 111, a standardized value based on, for example, 0 is registered. Standardization is an operation in which the value of each material parameter is subtracted from the average value of the entire database of each material parameter and divided by the standard deviation, resulting in a standard deviation with an average of 0 and a variance of 1. In this way, each material is quantified by the five sense parameters (material parameters). In the example of Figure 4, the value of the five sense parameter "hardness" for product number "101a001b001" is "0.12", and the value of the five sense parameter "warm/cold" is "0.03".

Depending on the product type, there are materials that are not affected by some of the five senses (such as hearing or taste). For example, taste may slightly affect the impression of food packaging materials, but has no effect on the impression of building materials for homes or synthetic leather materials for bags.

The emotion word parameter extraction unit 132 acquires emotion words output from the emotion word input unit 122. As shown in an example in FIG. 2, information words are words that express human emotions. The emotion word parameter extraction unit 132 also refers to emotion word parameter information 112 stored in the parameter information storage unit 110. The emotion word parameter extraction unit 132 extracts the value of an emotion word parameter associated with the acquired emotion word from the emotion word parameter information 112.

Emotion word parameter information 112 is information that indicates the degree of association between emotion words and five sense parameters. As mentioned above, the five sense parameters are composed of items that indicate sensations that are further subdivided into each of the five senses, for example, as shown in FIG. 3.

FIG. 5 is a diagram showing an example of emotion word parameter information 112 stored in the parameter information storage unit 110 of the material recommendation device 100 in one embodiment of the present disclosure. The emotion word parameter information 112 is data indicating the association between emotion words and the five sense parameters described above. Hereinafter, the five sense parameters associated with emotion words are referred to as "emotion word parameters."

In the emotion word parameter information 112 shown in FIG. 5, emotion words and emotion word parameters are associated with each other. Each row of the emotion word parameter information 112 is information about one emotion word. In the example of FIG. 5, the value of the five sense parameter "texture 1" for the emotion word "beautiful" is "0.73", and the value of the five sense parameter "texture 2" is "-0.15".

In this way, each emotion word is quantified in advance using five sense parameters (emotion word parameters). One possible method for quantifying the emotion words is to perform a sensory evaluation of each ingredient using emotion words on multiple subjects in advance, and then create a regression model between the emotion words and the five sense parameters to determine the numerical value. Furthermore, as with the ingredient parameters, these emotion word parameters may use standardized values using the entire emotion word database.

For each emotion word parameter in the emotion word parameter information 112, a standardized value based on 0 is registered, similar to the values of the material parameters shown in FIG. 4. Standardization is an operation in which the value of each emotion word parameter is subtracted from the average value of the entire database of emotion word parameters and divided by the standard deviation, resulting in a standard deviation with an average of 0 and a variance of 1. In this way, each emotion word is quantified using the five sense parameters (emotion word parameters).

The modal correction unit 133 acquires the usage information output from the usage input unit 123. As described above, the usage information is information that specifies the situation in which the user is impressed and according to which of the five senses. In addition, the modal correction unit 133 refers to the modal correction information 113 stored in the parameter information storage unit 110. The modal correction unit 133 corrects the values of the material parameters extracted by the material parameter extraction unit 131 and the values of the emotion word parameters extracted by the emotion word parameter extraction unit 132 based on the acquired usage information and the modal correction information 113.

Specifically, for example, when a user receives impressions through multiple of the five senses, the modal correction unit 133 identifies, based on the usage situation information, the order in which the impressions are received through each sense. The modal correction unit 133 extracts, from the modal correction information 113, a set of coefficients for taking into account the cross-modal effect occurring in the identified usage situation. The modal correction unit 133 multiplies the extracted set of coefficients by the values of the material parameters extracted by the material parameter extraction unit 131 and the values of the emotion word parameters extracted by the emotion word parameter extraction unit 132, respectively.

Generally, a cross-modal effect refers to a relationship between impressions received through two of the five senses, such as vision and touch. Generally, a multi-modal effect refers to a relationship between impressions received through multiple senses, such as vision, touch, and hearing. However, in this embodiment, the cross-modal effect also includes the meaning of the multi-modal effect.

For example, when a user recognizes a bag, they may look at it visually and then touch it to check its texture, or conversely, they may touch it to check its texture and then look at it visually. The overall impression that the user gets from a material will differ depending on whether the impression is received first by the sense of sight or touch. For example, materials such as synthetic leather for bags give the user different impressions depending on the order in which multiple senses are activated, creating a cross-modal effect.

The material recommendation device 100 in this embodiment takes into account the cross-modal effect according to the usage situation based on the acquired usage situation information. Specifically, as described above, the modal correction unit 133 multiplies the values of the material parameters extracted by the material parameter extraction unit 131 and the values of the emotion word parameters extracted by the emotion word parameter extraction unit 132 by a set of coefficients for taking the cross-modal effect into account. As a result, the values of the material parameters and the emotion word parameters are corrected to values that reflect the influence of the cross-modal effect according to the usage situation information.

In this embodiment, modal correction information 113 including a group of coefficients to be multiplied by the material parameter values (for example, modal correction information 113 shown in FIG. 6) and modal correction information 113 including a group of coefficients to be multiplied by the emotional word parameter values (for example, modal correction information 113 shown in FIG. 7) are pre-stored separately in the parameter information storage unit 110.

FIGS. 6 and 7 are diagrams showing an example of modal correction information 113 stored in the parameter information storage unit 110 of the material recommendation device 100 in one embodiment of the present disclosure.

Figure 6 shows an example of modal correction information 113 including a group of coefficients by which the material parameter values are multiplied. Figure 6 shows a group of coefficients corresponding to each piece of usage information, which are used to correct the material parameter values. As shown in Figure 6, in the case of coefficients by which the material parameter values are multiplied, for example, in cross-modal, the coefficient value of the sensation received first is preset to 1, the coefficient value of the sensation received later is preset to a value greater than 0 and less than 1, and the coefficient values of the other sensations are preset to 0.

As a method for quantifying such coefficients, similar to the above-mentioned case where emotional words are quantified in advance using the five sense parameters, a method can be considered in which a sensory evaluation is carried out in advance on multiple subjects using each material for each usage situation, and a regression model between the usage situation and the material is created to determine the numerical value.

FIG. 7 shows an example of modal correction information 113 including a group of coefficients by which the value of the emotion word parameter is multiplied. FIG. 7 shows a group of coefficients corresponding to each piece of usage information, which are used to correct the value of the emotion word parameter. As shown in FIG. 7, in the case of coefficients by which the value of the emotion word parameter is multiplied, for example, the coefficient value of the sensation that produces a cross-modal effect is preset to 1, and the coefficient values of the other sensations are preset to 0.

The material parameter extraction unit 131 outputs information indicating the material parameters corrected by the modal correction unit 133 to the recommended material determination unit 134. In addition, the emotion word parameter extraction unit 132 outputs information indicating the emotion word parameters corrected by the modal correction unit 133 to the recommended material determination unit 134.

The recommended material determination unit 134 acquires information indicating the material parameters output from the material parameter extraction unit 131 and information indicating the emotion word parameters output from the emotion word parameter extraction unit 132. Based on the acquired information, the recommended material determination unit 134 calculates the distance between the coordinates of the material parameter value and the emotion word parameter value for each material. The recommended material determination unit 134 determines the material to be recommended based on the calculated distance. For example, the recommended material determination unit 134 determines the material with the shortest calculated distance, or a predetermined number of materials in order of shortest calculated distance, as the recommended materials.

The recommended material determination unit 134 outputs information indicating the recommended material to the output unit 140. Note that the recommended material determination unit 134 may also output information indicating the distance between the material parameter value and the emotional word parameter value for each recommended material.

Here, a method of simply comparing the distance between the material parameter value and the emotion parameter value is used to select materials that match the emotion words, but other methods, such as a method of solving an inverse problem such as Bayesian optimization, or a method of solving a forward problem that minimizes error by performing a grid search, may be used. It is also possible to configure a system to perform the above tasks by using AutoML (Automated Machine Learning) such as DataRobot (registered trademark).

The output unit 140 is configured using an output device that presents information to the user. The output device here is, for example, a display device such as a liquid crystal display (LCD), a cathode ray tube (CRT), or an organic electro-luminescence (EL) display. The output unit 140 may also be a communication interface that transmits information to an external device that presents the information to the user. The external device is, for example, an information processing device such as a general-purpose computer, various displays, or a printer.

The output unit 140 acquires the information indicating the recommended materials output from the recommended material determination unit 134. The output unit 140 presents the information indicating the recommended materials to the user.

[Operation of the material recommendation device]
An example of the operation of the material recommendation device 100 according to the present disclosure will be described below with reference to a specific example. Fig. 8 is a flowchart showing the operation of the material recommendation device 100 according to an embodiment of the present disclosure.

First, the product type input unit 121 accepts input of product type information by the user. For example, a list of values for the "product type" item, which is the leftmost column of the material parameter information 111 shown in FIG. 4, ("food packaging material," "residential building material," "synthetic leather for bags," ...) is displayed on a display device such as a display. The user performs an input operation to select the desired product type from the displayed list using an input device such as a mouse. Note that, as an example, it is assumed here that the user has selected the product type of "synthetic leather for bags."

The product type input unit 121 outputs product type information indicating "synthetic leather material for bags" to the material parameter extraction unit 131. The material parameter extraction unit 131 acquires the product type information output from the product type input unit 121 (step S01).

Next, the emotion word input unit 122 accepts the input of emotion words by the user. For example, a list of emotion words as shown in FIG. 2 ("favorite," "flashy," "adorable," ...) is displayed on a display device such as a display. The user performs an input operation to select a desired emotion word from the displayed list using an input device such as a mouse. Note that, as an example, the user has selected the emotion word "beautiful." Note that this is not a limitation to this example, and multiple emotion words may be selected. In this case, the following process is executed for each emotion word.

The emotion word input unit 122 outputs the emotion word "beautiful" to the emotion word parameter extraction unit 132. The emotion word parameter extraction unit 132 acquires the emotion word output from the emotion word input unit 122 (step S02).

Next, the usage status input unit 123 accepts the input of usage status information by the user. For example, a list of values for the "first" and "last" items of "cross-modal effect" ("tactile", "none", "visual", "none", "auditory", "none", ...) in the second column from the left of the modal correction information 113 shown in FIG. 6 or 7 is displayed on a display device such as a display. The user imagines the usage situation when using the product and performs an input operation to select the desired usage status information from the displayed list using an input device such as a mouse. Note that, as an example, it is assumed here that the user selects usage status information in which the first sensation received is "visual" and the second sensation received is "tactile".

The usage status input unit 123 outputs usage status information indicating that the first sensation is "visual" and the second sensation is "tactile" to the modal correction unit 133. The modal correction unit 133 acquires the usage status information output from the usage status input unit 123 (step S03).

Note that a list of options, such as "Decide whether to purchase just by looking at the bag" or "Decide whether to purchase after checking the feel of the bag after looking at it" may be displayed on a display device such as a display, and the user may select from the list. In this case, for example, the parameter information storage unit 110 stores information in advance that associates these options with one or more senses. That is, for example, information in advance is stored in which "Decide whether to purchase just by looking at the bag" is associated with "sight," and "Decide whether to purchase after checking the feel of the bag after looking at it" is associated with "sight" and "touch."

Then, when the modal correction unit 133 obtains information from the usage status input unit 123 indicating that the option "After looking at the bag, check the feel to decide whether to purchase it or not" has been selected by the user, the modal correction unit 133 can determine by referring to the above information stored in advance in the parameter information storage unit 110 that the usage status information selected is one in which the first sensation received is "sight" and the second sensation received is "touch."

Then, the material parameter extraction unit 131 extracts the values of the material parameters associated with the product type corresponding to the acquired product type information from the material parameter information 111 pre-stored in the parameter information storage unit 110 (step S04). For example, since the product type "synthetic leather for bags" has been selected by the user here, the material parameter extraction unit 131 extracts all the data (values of the material parameters) in the rows in which the value of the "product type" item in the material parameter information 111 shown in FIG. 4 is "synthetic leather for bags".

In the material parameter information 111 shown in FIG. 4, there are five rows in which the value of the "Product Type" item is "Synthetic Leather for Bags" (m=5). That is, a total of five materials used as synthetic leather for bags are registered (product numbers "301a201b101" to "305a201b105"). If the number of material parameters (sensory parameters) associated with each material is n, then the material parameter extraction unit 131 will extract m (5) matrices consisting of 1 row and n columns (hereinafter referred to as "Matrix A").

Next, the modal correction unit 133 corrects the values of the material parameters extracted by the material parameter extraction unit 131 based on the acquired usage information and the modal correction information 113 (step S05).

For example, here, the usage status information selected by the user is one in which the first sensation is "visual" and the second sensation is "tactile". Therefore, the modal correction unit 133 extracts data (a group of coefficients) from a row in which the value of the "first" item in the "cross-modal effect" item of the modal correction information 113 shown in FIG. 6 is "visual" and the value of the "last" item is "tactile". If the number of material parameters associated with each material is n, the modal correction unit 133 extracts one matrix consisting of 1 row and n columns. The modal correction unit 133 then generates a diagonal matrix consisting of n rows and n columns (hereinafter referred to as "diagonal matrix B") with the extracted matrix as its diagonal components.

The modal correction unit 133 corrects the values of the material parameters by multiplying each of the extracted matrices A by the generated diagonal matrix B. This causes the values of the material parameters to be corrected to values that take into account the cross-modal effects according to the selected usage information.

FIG. 9 is a diagram showing an example of a material parameter value correction process performed by the modal correction unit 133 of the material recommendation device 100 in one embodiment of the present disclosure. As an example, FIG. 9 shows a matrix calculation when correcting the material parameter values of a synthetic leather material for a bag that is assigned the product number "301a201b101". As shown in FIG. 9, a matrix consisting of corrected material parameter values is calculated by multiplying matrix A consisting of material parameter values extracted from the material parameter information 111 shown in FIG. 4 by diagonal matrix B generated from a group of coefficients extracted from the modal correction information 113 shown in FIG. 6 ("-0.09", "0.71", "0.69", "0.26", ...).

Next, the emotion word parameter extraction unit 132 extracts the value of the emotion word parameter associated with the acquired emotion word from the emotion word parameter information 112 pre-stored in the parameter information storage unit 110 (step S06). For example, since the emotion word "beautiful" has been selected by the user here, the emotion word parameter extraction unit 132 extracts data (emotion word parameter value) from the row in which the value of the "emotion word" item in the emotion word parameter information 112 shown in FIG. 5 is "beautiful."

If the number of emotion word parameters (sensory parameters) associated with each emotion word is n, then the emotion word parameter extraction unit 132 extracts m matrices (hereinafter referred to as "matrix C") consisting of one row and n columns. The value of m here represents the number of emotion words selected by the user. Here, as an example, since only one emotion word ("beautiful") has been selected by the user, the emotion word parameter extraction unit 132 extracts one matrix C.

Next, the modal correction unit 133 corrects the values of the emotion word parameters extracted by the emotion word parameter extraction unit 132 based on the acquired usage information and the modal correction information 113 (step S07).

For example, here, the usage information selected by the user is one in which the first sensation is "visual" and the second sensation is "tactile". For this reason, the modal correction unit 133 extracts data (a group of coefficients) from rows in which the value of the "first" item in the "cross-modal effect" item of the modal correction information 113 shown in FIG. 7 is "visual" and the value of the "last" item is "tactile". If the number of emotion word parameters associated with each emotion word is n, the modal correction unit 133 extracts one matrix consisting of 1 row and n columns. The modal correction unit 133 then generates a diagonal matrix consisting of n rows and n columns (hereinafter referred to as "diagonal matrix D") with the extracted matrix as its diagonal components.

The modal correction unit 133 corrects the values of the emotion word parameters by multiplying each of the extracted matrices C (here, one) by the generated diagonal matrix D. In this way, the emotion word parameter values are corrected to values that take into account the cross-modal effect.

FIG. 10 is a diagram showing an example of the correction process of the emotion word parameter values performed by the modal correction unit 133 of the material recommendation device 100 in one embodiment of the present disclosure. FIG. 10 shows a matrix calculation when correcting the emotion word parameter value corresponding to the emotion word "beautiful". As shown in FIG. 10, a matrix C consisting of the emotion word parameter values extracted from the material parameter information 111 shown in FIG. 5 is multiplied by a diagonal matrix D generated from a group of coefficients extracted from the modal correction information 113 shown in FIG. 7 to calculate a matrix consisting of the corrected emotion word parameter values ("0.03", "-0.81", "-0.62", "0.03", ...).

The material parameter extraction unit 131 outputs the corrected material parameter values to the recommended material determination unit 134. In addition, the emotion word parameter extraction unit 132 outputs the corrected emotion word parameter values to the recommended material determination unit 134.

Next, the recommended material determination unit 134 obtains the corrected material parameter values output from the material parameter extraction unit 131 and the corrected emotion word parameter values output from the emotion word parameter extraction unit 132. The recommended material determination unit 134 regards the obtained material parameter matrix and emotion word parameter matrix as coordinates, and calculates the distance between the coordinates for each material. The recommended material determination unit 134 determines the material to be recommended based on the calculated distance (step S08). For example, from among multiple materials, materials with a short distance between the coordinates of the emotion word parameter corresponding to the emotion word specified by the user and the coordinates of the material parameter corresponding to the material can be regarded as having a strong relationship with the emotion word, and can be identified in order.

FIG. 11 is a diagram showing an example of a calculation process of the distance between coordinates performed by the recommended material determination unit 134 of the material recommendation device 100 in one embodiment of the present disclosure. FIG. 11 shows a process of subtracting the components of a matrix consisting of the corrected material parameter values calculated by the correction process shown in FIG. 9 from the matrix consisting of the corrected emotion word parameter values calculated by the correction process shown in FIG. 10, squaring each value of the calculated matrix components, and further adding up the calculated values to perform a root calculation to calculate the distance between coordinates.

For example, in FIG. 11, the value of the emotion word parameter for "hardness" is "0.03", and the value of the material parameter for "hardness" is "-0.09", so subtracting and squaring it results in "0.0144". The recommended material determination unit 134 adds up the component values calculated for each of the five sense parameters in this way, and further performs a root calculation to calculate the value, which is the distance between the coordinates of the emotion word parameter and the material parameter. Note that here, the calculated distance between the coordinates is, for example, the square root of "4.0938 + α". However, the calculation of the distance between the coordinates is not limited to this example, and may be calculated by other methods known in the technical field.

The recommended material determination unit 134 performs the above calculation for each material and each emotion word, and calculates the distance between the coordinates. The recommended material determination unit 134 determines the material to be recommended based on the calculated distance between the coordinates. The recommended material determination unit 134 outputs information indicating the recommended material and information indicating the calculated distance between the coordinates for the material to the output unit 140.

The output unit 140 acquires information indicating the recommended material output from the recommended material determination unit and information indicating the distance between the coordinates calculated for the material. The output unit 140 presents the information indicating the recommended material and information indicating the distance between the coordinates calculated for the material to the user (step S09).

12 is a diagram showing an example of information output by the output unit 140 of the material recommendation device 100 in one embodiment of the present disclosure. As shown in FIG. 12, the information output by the output unit 140 includes, for example, the selected product type ("artificial leather for bags"), the selected emotion word ("beautiful"), and the value of the distance between the coordinates for each material (for each product number). At this time, the output unit 140 outputs the materials in order of the shortest distance between the coordinates, so that the more recommended materials are displayed at the top. Note that, as shown in FIG. 12, the distance between the coordinates is also displayed as a bar graph, allowing the user to more intuitively recognize the recommendation level of each material. However, this is not limited to this example, and the distance between the coordinates may be displayed in a predetermined order of the product numbers.

The following is an example of a method for presenting recommended materials by the output unit 140 when the user has selected multiple emotion words. Figure 13 is a diagram showing another example of information output by the output unit 140 of the material recommendation device 100 in one embodiment of the present disclosure. Here, it is assumed that the user has selected the emotion words "flashy" and "manly."

In the example of FIG. 13, the information output by the output unit 140 includes, for example, the selected product type ("synthetic leather for bags"). The information output by the output unit 140 also includes, for example, the value of the distance between the coordinates for each material (for each product number) for each selected emotion word ("flashy" and "manly"). The information output by the output unit 140 also includes, for example, the average value of the distance between the coordinates for multiple emotion words for each material.

By presenting information in this manner by the output unit 140, the user can recognize that the synthetic leather material for bags that is closest to the impression of "flashy" is the material assigned with the product number "303a201b102" with a coordinate distance of "8.7". The user can also recognize that the synthetic leather material for bags that is closest to the impression of "masculine" is the material assigned with the product number "303a201b104" with a coordinate distance of "8.3". The user can also recognize that the synthetic leather material for bags that has both the impression of "flashy" and the impression of "masculine" that are high on average is the material assigned with the product number "303a201b104" with a coordinate distance of "9.4". When three or more emotion words are specified, the average value of the three emotion words may be output, or the average value may be output for each combination of two of the three emotion words.

This allows the user to, for example, select the material with the shortest distance between all coordinates, or to select the material that has the closest impression on average to the multiple emotion words selected.

For example, the output unit 140 may generate and present a map that allows visual understanding of the coordinates of the material parameters and the coordinates of the emotion word parameters. Also, for example, the output unit 140 may perform analysis using principal component analysis or multidimensional scaling on the coordinates of the material parameters and the coordinates of the emotion word parameters, and visually present the analysis results. For example, in the case of principal component analysis, the output unit 140 may present the positional relationship between the coordinates of the material parameters and the emotion word parameters using two axes, the first principal component and the second principal component, in a two-dimensional diagram using BiPlot.

As described above, the material recommendation device 100 in one embodiment of the present disclosure includes an input unit 120, a material parameter extraction unit 131, an emotion word parameter extraction unit 132, and a recommended material determination unit 134. The input unit 120 accepts input of information indicating a product type and an emotion word indicating a word expressing an emotion. The material parameter extraction unit 131 extracts a sensory parameter (material parameter) value for each material corresponding to the product type input to the input unit 120 based on material parameter information in which materials used in products belonging to the product type are associated with sensory parameter values indicating the degree of sensation humans experience from the material. The emotion word parameter extraction unit 132 extracts a sensory parameter (emotion word parameter) value corresponding to the emotion word input to the input unit 120 based on emotion word parameter information in which emotion words are associated with sensory parameter values. The recommended ingredient determination unit 134 determines ingredients recommended for the emotion word input to the input unit 120 based on the sensory parameter values for each ingredient extracted by the ingredient parameter extraction unit 131 and the sensory parameter values extracted by the emotion word parameter extraction unit 132.

With this configuration, the material recommendation device 100 can automatically recommend appropriate materials according to a person's emotions, without relying on the sense of a specific person who recommends the materials.

Furthermore, the material recommendation device 100 corrects the sensory parameter values for each material and the sensory parameter values corresponding to emotion words to take into account the influence of cross-modal effects. With this configuration, the material recommendation device 100 can more accurately recommend appropriate materials according to a person's emotions.

(Modification)
Of the functional configuration of the material recommendation device 100 shown in FIG. 1, the modal correction information 113, the usage status input unit 123, and the modal correction unit 133 may be omitted to form a simpler configuration. In other words, it is also possible to form a material recommendation device from the functions of the material recommendation device 100, omitting the configuration for correcting the five sense parameters in consideration of the cross-modal effect. The material recommendation device 100 may determine a recommended material without correcting the sensory parameter values for each material and the sensory parameter values according to the usage status information. In this way, by making the configuration simpler, it is possible to reduce the device cost and operation cost of the material recommendation device 100, and the man-hours required to prepare the modal correction information 113, etc.

The above-mentioned processing may be performed by recording a program for implementing the functions of the material recommendation device 100 in the embodiment on a computer-readable recording medium, and having the computer system read and execute the program recorded on the recording medium. Note that the term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer system" also includes a WWW system equipped with a homepage providing environment (or display environment). The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CD-ROMs, and storage devices such as hard disks built into a computer system. The term "computer-readable recording medium" also refers to storage devices that hold a program for a certain period of time, such as volatile memory (RAM) inside a computer system that becomes a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line.

The above program may also be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The above program may also be one that realizes part of the above-mentioned functions. Furthermore, it may be a so-called difference file (difference program) that can realize the above-mentioned functions in combination with a program already recorded in the computer system.

Reference Signs List 100: Material recommendation device 110: Parameter information storage unit 111: Material parameter information 112: Emotion word parameter information 113: Modal correction information 120: Input unit 121: Product type input unit 122: Emotion word input unit 123: Usage status input unit 131: Material parameter extraction unit 132: Emotion word parameter extraction unit 133: Modal correction unit 134: Recommended material determination unit 140: Output unit

Claims (9)

1. an input unit that receives input of information indicating a product type and an emotion word indicating a word expressing an emotion;
   a material parameter extraction unit that extracts a sensory parameter value for each material corresponding to the product type input to the input unit, based on material parameter information in which materials used in the product belonging to the product type are associated with sensory parameter values indicating the degree of sensation experienced by humans with respect to the material;
   an emotion word parameter extraction unit that extracts a value of a sensory parameter corresponding to the emotion word input to the input unit based on emotion word parameter information in which the emotion word and the value of the sensory parameter are associated with each other;
   a recommended ingredient determination unit that determines the ingredient recommended for the emotion word input to the input unit, based on the sensory parameter values for each ingredient extracted by the ingredient parameter extraction unit and the sensory parameter values extracted by the emotion word parameter extraction unit;
   A material recommendation device comprising:
2. The material recommendation device according to claim 1 , wherein the emotion words are words classified as evaluation factors or activity factors among factors of an EPA (Evaluation, Potency, Activity) structure defined in impression evaluation by an SD (Semantic Differential) method.
3. The ingredient recommendation device according to claim 1 or 2, further comprising a correction unit that corrects the values of the sensory parameters using coefficients that take into account a cross-modal effect in which impressions received through a plurality of types of senses are related to each other.
4. The coefficients are set to different values according to the order in which impressions are received by the plurality of types of senses,
   The material recommendation device according to claim 3 , wherein the correction unit acquires information indicating the order, and performs correction using the coefficient corresponding to the acquired order.
5. The material recommendation device according to claim 4, wherein the coefficient used to correct the value of the sensory parameter for each material is set to a value of 1 for the sensation received first, a value greater than 0 and less than 1 for the sensation received later, and a value of 0 for any other sensation.
6. The material recommendation device according to claim 4, wherein the coefficients used to correct the values of the emotion word sensory parameters are set to a value of 1 for the sensory parameters that cause the cross-modal effect, and are set to a value of 0 for the sensory parameters that do not cause the cross-modal effect.
7. The material recommendation device according to claim 1 or 2, wherein the senses are the five senses or senses further subdivided from the five senses.
8. an input step of receiving an input of information indicating a product type and an emotion word indicating a word expressing an emotion;
   a material parameter extraction step of extracting a sensory parameter value for each material corresponding to the product type input in the input step, based on material parameter information in which materials used in the product belonging to the product type are associated with sensory parameter values indicating the degree of sensation experienced by humans with respect to the material;
   an emotion word parameter extraction step of extracting a value of a sensory parameter corresponding to the emotion word input in the input step, based on emotion word parameter information in which the emotion word and the value of the sensory parameter are associated with each other;
   a recommended ingredient determination step of determining the ingredient recommended for the emotion word input in the input step, based on the sensory parameter values for each ingredient extracted in the ingredient parameter extraction step and the sensory parameter values extracted in the emotion word parameter extraction step;
   A material recommendation method having:
9. On the computer,
   an input step of receiving an input of information indicating a product type and an emotion word indicating a word expressing an emotion;
   a material parameter extraction step of extracting a sensory parameter value for each material corresponding to the product type input in the input step, based on material parameter information in which materials used in the product belonging to the product type are associated with sensory parameter values indicating the degree of sensation experienced by humans with respect to the material;
   an emotion word parameter extraction step of extracting a value of a sensory parameter corresponding to the emotion word input in the input step, based on emotion word parameter information in which the emotion word and the value of the sensory parameter are associated with each other;
   a recommended ingredient determination step of determining the ingredient recommended for the emotion word input in the input step, based on the sensory parameter values for each ingredient extracted in the ingredient parameter extraction step and the sensory parameter values extracted in the emotion word parameter extraction step;
   A program for executing.

Images