Food information providing method, food information providing system, food information providing device, food information providing program, and storage medium recording food information providing program

Abstract

To provide a food information providing method that can provide food information while giving an enough convinced feeling.SOLUTION: A food information providing method to be implemented in a providing system S, comprises: a user information recognition unit 10 that recognizes user information; a selection reason recognition unit 13 that inputs the user information, and outputs a selection reason, using a first prediction model outputting the selection reason being a reason to adopt upon selecting a food in which a user performs an eating behavior; a recommendation food recognition unit 14 that recognizes a recommendation food; and a food information providing unit 16 that provides a user with food information including the selection reason corresponding to the recommendation food. The first prediction model is configured to, with each of a plurality of candidate reasons and the user information as learning data, machine-learn a correlation between the candidate reason and the user information, and select the selection reason from the plurality of candidate reasons, and output the selected selection reason.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to a cooking information providing system, a cooking information providing method, a cooking information providing device, a cooking information providing program, and a storage medium having the program recorded thereon, which provide a user with information regarding dishes for which eating behavior is recommended. [Background technology]

[0002] Conventionally, there is a system that presents recommended dishes (hereinafter, referred to as "recommended dishes") to a user as a recommended eating behavior. Here, "eating behavior" refers to various behaviors related to food intake, including not only the act of eating a dish, but also the act of cooking and preparing the dish, and making a reservation at a restaurant that serves the dish.

[0003] Incidentally, recommended dishes differ not only depending on the user's personal conditions, such as the user's preferences for food, but also on environmental conditions (e.g., the amount of time available for eating, the ingredients available, etc.). As such, a system that presents recommended dishes taking into account the user's preferences for food as well as the situation (i.e., environmental conditions) is known (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] International Publication No. 2020 / 241693 Summary of the Invention [Problem to be solved by the invention]

[0005] In order for the system described in Patent Document 1 to recognize recommended dishes, the user needs to input environmental conditions. Therefore, when dish information about recommended dishes is provided by this system, the user will assume that the dish information has been provided taking into account the user's own environmental conditions.

[0006] However, there are cases where the user does not have the knowledge to understand the relationship between the recommended dishes and environmental conditions. In addition, there are cases where the user does not accurately understand his / her own environmental conditions. In such cases, the user may not be fully convinced that the recommended dishes were recommended (and therefore that the dish information was provided).

[0007] The present invention has been made in consideration of the above points, and aims to provide a cooking information providing method, a cooking information providing system, a cooking information providing device, a cooking information providing program, and a storage medium having the program recorded thereon, which are capable of providing cooking information while providing a sufficient sense of convincingness. [Means for solving the problem]

[0008] In the following, the term "user" refers to a user of the cooking information providing system, cooking information providing method, cooking information providing device, cooking information providing program, and storage medium on which the program is recorded according to the present invention. Therefore, users include not only the person who performs the eating behavior, but also a person who makes others perform the eating behavior, and a person who is made to perform the eating behavior by others. For example, users include a parent who uses the system for their child, and a child whose eating behavior is managed by a parent.

[0009] In addition, "food information" includes information about the food itself, such as the name of the food, ingredients, and recipe, as well as information about the store where the food is served and information about the means of obtaining the food, such as the names of services that can deliver the food.

[0010] Moreover, the term "learning data" refers to data used to train a machine learning algorithm, and examples of such data include feature values, correct answer data, or combinations thereof used in the training. More specifically, the term "learning data" refers to data used to determine parameters possessed by a machine learning algorithm, and examples of such data include feature values, correct answer data, or combinations thereof used to determine the parameters.

[0011] The method for providing recipe information of the present invention comprises the steps of: A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit inputting the user information and outputting selection reasons, which are reasons influencing the user's selection of a dish, using a first prediction model; and a recommended dish recognition unit recognizing the recommended dishes; a cooking information providing unit providing the user with the cooking information including the selection reason corresponding to the recommended cooking dish; The first prediction model uses each of a plurality of candidate reasons that may be the reason for selection and the user information as learning data, performs machine learning to learn the correlation between the candidate reasons and the user information, and selects and outputs the reason for selection from the plurality of candidate reasons.

[0012] In this way, the method of providing recipe information of the present invention provides recipe information about recommended recipes to a user in a form that includes the selection reasons, which are reasons that influence the user's choice of recipe, that correspond to the recommended recipe. In other words, the selection reasons are provided as grounds for recommending the recommended recipe.

[0013] Here, the selection reason is output by converting user information (e.g., the user's environmental conditions, etc.) into a form of reason using a first prediction model that performs machine learning on the correlation between each of a plurality of candidate reasons that are possible selection reasons and the user information. This allows the user to directly grasp the reason for the recommendation of the recommended dish, compared to a case where the user information is provided as is.

[0014] Therefore, according to the cooking information providing method of the present invention, the reasons for selection suitable for each individual user are directly provided in the cooking information regarding the recommended dishes, so that the cooking information regarding the recommended dishes can be provided to the user in a format that gives a sufficient sense of convincing.

[0015] In addition, in the method for providing cooking information of the present invention, a current emotion recognition unit that recognizes a current emotion that is an emotion at a time when the user desires to acquire the recipe information; It is preferable that the first prediction model uses each of the plurality of candidate reasons, the user information, and the current emotion as learning data, performs machine learning to learn correlations between the candidate reasons and the user information and the current emotion, and selects and outputs the reason for selection from the plurality of candidate reasons.

[0016] In general, even if the environment is the same, if the user's emotions are different, the impact of that environment on the user will be different. It is also known that there is a certain correlation between emotions and eating behavior (see, for example, Imada Sumio (2009). Emotions and Eating Behavior, Emotion Psychology Research 17, 120-128).

[0017] Therefore, by using the user's current emotions as well as user information such as the user's environment as input data used in the first prediction model, the selection reasons output by the first prediction model become even more suited to individual users.

[0018] In addition, in the method for providing cooking information of the present invention, When the current emotion is used as input data in the first prediction model, the current emotion recognition unit has an emotion model presenting unit that presents to the user an emotion model, which is a planar or three-dimensional model whose coordinates are defined based on a plurality of basic emotions, in a form in which any coordinate in the emotion model can be selected, and a coordinate recognition unit that recognizes selected coordinates, which are coordinates in the emotion model selected by the user, as the current emotion; It is preferable that the first prediction model uses each of the multiple candidate reasons, the user information, and the selected coordinates as learning data, performs machine learning to learn correlations between the candidate reasons and the user information and the selected coordinates, and selects and outputs the selection reason from the multiple candidate reasons.

[0019] In general, it is difficult for a user to accurately express his / her own emotions (and thus to input them into a system that executes a method), and the user may feel troublesome if he / she makes efforts to accurately express his / her own emotions, which may result in further changes in emotions.

[0020] Thus, it is advisable to adopt an emotion model, which is a planar or three-dimensional model whose coordinates are defined based on a plurality of basic emotions, as an interface for inputting the user's emotions, and configure the interface so that the coordinates selected in the emotion model are recognized as the user's emotions.

[0021] This allows the user to easily and accurately express his / her own feelings as coordinate values ​​without converting them into linguistic expressions. Furthermore, since the coordinate parameter is clearer than the emotion parameter, using the coordinate parameter as input data eliminates ambiguity from the input data. As a result, the selection reason, which is output data, can be made more suitable for each individual user. In turn, the selection reason can give the user a higher sense of satisfaction.

[0022] In addition, in the method for providing cooking information of the present invention, When adopting an emotion model as an interface for inputting user emotions, The emotion model is preferably a model including a first coordinate axis representing one or more basic emotions, and a second coordinate axis representing the strength of the basic emotions.

[0023] If a model using such coordinate axes is adopted as an emotion model, users can intuitively understand what emotion the coordinates in the emotion model indicate. This in turn makes it easier for users to accurately express their own emotions as coordinates, further eliminating ambiguity from input data.

[0024] As emotion models using such coordinate axes, for example, EmojiGrid (registered trademark), Plutchik's Wheel of Emotions, and models generated based on them (for example, a circular model that uses the color arrangement of Plutchik's Wheel of Emotions) can be adopted. In addition, as basic emotions, for example, the emotions adopted in Plutchik's Wheel of Emotions, such as joy, trust, fear, surprise, sadness, disgust, anger, and anticipation, and the emotions adopted in Ekman's theory, such as joy, fear, surprise, sadness, disgust, and anger, can be adopted.

[0025] In addition, in the method for providing cooking information of the present invention, the recommended dish recognition unit recognizes the recommended dish using a second prediction model that inputs the user information, the current emotion, and the selection reason to output the recommended dish; It is preferable that the second prediction model uses each of a plurality of candidate dishes that may become the recommended dish, as well as the user and the reason for selection as learning data, and performs machine learning to detect correlations between the candidate dishes and the user information and the reason for selection, and selects and outputs the recommended dish from the plurality of candidate dishes.

[0026] In this way, by adopting a prediction model that uses not only user information but also the selection reasons as input data as the second prediction model that outputs recommended dishes, the recommended dishes that are output can be not only suitable for individual users, but also dishes that are optimized for various situations (and therefore diverse). This allows the user to be highly convinced about the recommended dishes themselves.

[0027] The method for providing cooking information of the present invention further comprises the steps of: A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit recognizing a selection reason, which is a reason that influences a selection of a dish by the user, based on correlation data representing a correlation between the user information and each of a plurality of candidate reasons; a recommended dish recognition unit recognizing the recommended dishes; The method further comprises a step of: a dish information providing unit providing the user with the dish information including the selection reason corresponding to the recommended dish.

[0028] In this way, the method of providing recipe information of the present invention provides recipe information about recommended recipes to a user in a form that includes the selection reasons, which are reasons that influence the user's choice of recipe, that correspond to the recommended recipe. In other words, the selection reasons are provided as grounds for recommending the recommended recipe.

[0029] Here, the selection reason is output by converting user information (e.g., the user's environmental conditions, etc.) into a form of reason based on correlation data that indicates the correlation between the user information and each of the multiple candidate reasons. This allows the user to directly grasp the reason for the recommendation of the recommended dish, compared to when the user information is provided as is.

[0030] Therefore, according to the cooking information providing method of the present invention, the reasons for selection suitable for each individual user are directly provided in the cooking information regarding the recommended dishes, so that the cooking information regarding the recommended dishes can be provided to the user in a format that gives a sufficient sense of convincing.

[0031] The method for providing cooking information of the present invention further comprises the steps of: A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit inputting the user information and outputting selection reasons, which are reasons influencing the user's selection of a dish, using a generative model to output the selection reasons; a recommended dish recognition unit recognizing the recommended dishes; a cooking information providing unit providing the user with the cooking information including the selection reason corresponding to the recommended cooking dish; The generative model uses first text data expressing reasons influencing the selection of a dish and the user information as learning data, performs machine learning to detect correlation between the first text data and the user information, and generates and outputs second text data expressing the reasons for selection.

[0032] In this way, the method of providing recipe information of the present invention provides recipe information about recommended recipes to a user in a form that includes the selection reasons, which are reasons that influence the user's choice of recipe, that correspond to the recommended recipe. In other words, the selection reasons are provided as grounds for recommending the recommended recipe.

[0033] Here, the selection reason is text data generated by referring to user information (e.g., the user's environmental conditions, etc.), which allows the user to directly grasp the reason for the recommendation of the recommended dish, compared to when the user information is provided as is.

[0034] Therefore, according to the cooking information providing method of the present invention, the reasons for selection suitable for each individual user are directly provided in the cooking information regarding the recommended dishes, so that the cooking information regarding the recommended dishes can be provided to the user in a format that gives a sufficient sense of convincing.

[0035] The cooking information providing system of the present invention comprises: A cooking information providing system that provides a user with cooking information that is information about recommended cooking, which is cooking that is recommended for a user to eat, comprising: The system includes a user information recognition unit, a selection reason recognition unit, a recommended dish recognition unit, and a dish information provision unit, The present invention is characterized in that it is configured to execute any one of the above-mentioned cooking information providing methods.

[0036] In addition, the cooking information providing device of the present invention comprises: A cooking information providing device that provides a user with cooking information that is information about recommended cooking, which is cooking that is recommended for a user to eat, comprising: The system includes a user information recognition unit, a selection reason recognition unit, a recommended dish recognition unit, and a dish information provision unit, The present invention is characterized in that it is configured to execute any one of the above-mentioned cooking information providing methods.

[0037] In addition, the cooking information providing program of the present invention is A recipe information providing method for providing a user with recipe information, the recipe information being information about recommended recipes that are recommended for a user's eating behavior, comprising: The present invention is characterized in that the computer is caused to execute any one of the above-mentioned methods for providing recipe information.

[0038] The recording medium of the present invention is The present invention is characterized in that the above-mentioned cooking information providing program is recorded and the cooking information providing program is readable by the computer. [Brief description of the drawings]

[0039] [Figure 1] FIG. 1 is an explanatory diagram showing a schematic configuration of a provision system according to an embodiment. [Diagram 2] FIG. 2 is a block diagram showing the configuration of a processing unit of the provision system of FIG. 1; [Figure 3A] FIG. 2 is a schematic diagram showing an example of an emotion model used to acquire a current emotion in the provision system of FIG. 1; [Figure 3B] FIG. 13 is a schematic diagram showing an example of an emotion model used to acquire a current emotion in the delivery system according to the first modified example. [Figure 3C] FIG. 13 is a schematic diagram showing an example of an emotion model used to acquire a current emotion in the delivery system according to the second modified example. [Figure 4] FIG. 13 is an image diagram showing an example of a data table used in a provision system according to a fourth modified example. [Diagram 5] 6 is a flowchart showing a process executed by the provision system of FIG. 1 when the system recognizes a reason for selection. [Figure 6] 4 is a flowchart showing the process executed by the provision system of FIG. 1 when providing dish information. [Figure 7] 10 is an image diagram showing an example of an image displayed on a user terminal when the providing system of FIG. 1 recognizes user information. [Figure 8] FIG. 2 is an image diagram showing an example of an image displayed on a user terminal when the providing system of FIG. 1 recognizes a current emotion. [Figure 9] 10 is an image diagram showing an example of an image displayed on a user terminal when the provision system of FIG. 1 provides dish information. [Figure 10] FIG. 2 is an image diagram showing an example of an image displayed on a user terminal when any of the dish information is selected by the provision system of FIG. 1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0040] A cooking information providing system (hereinafter, referred to as a "providing system S") according to an embodiment and a cooking information providing method executed using the same will be described below with reference to the drawings.

[0041] The provision system S in this embodiment is used in a service that recognizes recommended dishes, which are dishes that the user U is recommended to eat, in response to a request from the user U, extracts a selection reason for recommending the recommended dish from a number of candidate reasons, and provides the user U with dish information including the selection reason for the recommended dish.

[0042] In the following description, the term "user" refers to a user of the provision system S. Therefore, the person assumed to be a user U includes not only the person who performs eating behavior, but also a person who makes others perform eating behavior, and a person who is made to perform eating behavior by others. For example, it includes a parent who uses the system for his / her child, and a child whose eating behavior is managed by a parent.

[0043] Here, "food information" includes information about the food itself, such as the name of the food, ingredients, and recipe, as well as information about the store where the food is served and information about the means of obtaining the food, such as the names of services that can deliver the food.

[0044] [System Overview] Hereinafter, a schematic configuration of the provision system S will be described with reference to FIGS.

[0045] As shown in FIG. 1, the provision system S is a computer system for providing recipe information to a user U, and is composed of a server 1 owned by a provider of the service provided by the provision system S.

[0046] The server 1 is configured to be capable of communicating information with a user terminal 2 such as a smartphone or tablet owned by a user U via the Internet network, a public line, or the like.

[0047] The cooking information providing system of the present invention is not limited to being configured by one server, but any of the terminals constituting the cooking information providing system may be configured to include a processing unit as described below.

[0048] Therefore, for example, the entire cooking information providing system may be made up of multiple servers. Also, at least one of the processing units or at least part of the functions of that processing unit may be implemented in the user terminal, and the system may be configured with the user terminal and the server working together, or with the user terminal alone. Furthermore, in the providing system S, functions equivalent to the input unit 20 and output unit 21 provided in the user terminal 2 may be provided in a terminal having a processing unit, and the providing system S may be configured as an independent cooking information providing device.

[0049] Also, as shown in FIG. 2, a user terminal 2 capable of communicating with a server 1, which is a provision system S, has an input unit 20 and an output unit 21 as functions (processing units) realized by at least one of an implemented hardware configuration and a program.

[0050] In this embodiment, the input unit 20 and the output unit 21 of the user terminal 2 are described as being touch panels. However, the user terminal is not limited to such a configuration, and may be any device that can accept information input from a user and output information to provide the user. Therefore, for example, the user terminal may be configured to enable input and output using a keyboard, microphone, camera, speaker, etc. in addition to the touch panel.

[0051] [Configuration of each processing section] Next, the processing units constituting the estimation system S will be described with reference to FIGS.

[0052] As shown in FIG. 2, the server 1 includes a user information recognition unit 10, a user information storage unit 11, a current emotion recognition unit 12, a reason for selection recognition unit 13, a recommended dish recognition unit 14, a dish information storage unit 15 (dish information recognition unit), and a dish information provision unit 16, as functions (processing units) realized by at least one of the implemented hardware configuration and programs.

[0053] The user information recognition unit 10 recognizes information including at least one of the biometric information, preferences, and eating behavior environment of the user U as user information.

[0054] Here, "biometric information" refers to information about the user's body that may affect eating behavior. Examples of biometric information include the user's gender, age, height, and weight. Furthermore, here, "preferences" includes not only the user's preferences regarding cooking or ingredients themselves, but also the user's preferences that may affect eating behavior. For example, it includes whether or not the user likes summer (and thus the state of appetite depending on the temperature).

[0055] In addition, here, the "environment of eating behavior" includes, for example, the number of people with whom the user engages in eating behavior, the user's relationship with those with whom the eating behavior is conducted, the planned time for the eating behavior, motivation for the eating behavior, the place where the eating behavior is conducted, the type of eating behavior (whether the user eats the food themselves or feeds others, whether the user cooks the food or eats out, etc.), etc.

[0056] In the provision system S, the user information recognition unit 10 first presents a predetermined questionnaire to the output unit 21 of the user terminal 2 (see FIG. 7). After that, the user information recognition unit 10 refers to the answers to the questionnaire and the information stored in the user information storage unit 11, and recognizes user information related to the user U who will receive the service provided by the provision system S this time.

[0057] Furthermore, if the recognized user information differs from the user information stored in the user information storage unit 11, the user information recognition unit 10 stores new user information in the user information storage unit 11.

[0058] The user information storage unit 11 stores user information corresponding to each user U. First, the user information storage unit 11 stores, as initial information, user information acquired during a registration procedure for the user U to receive a service from the provision system S. Thereafter, when new user information is recognized by the user information recognition unit 10, information that has been added or changed based on the new user information is stored in the initial information.

[0059] In addition, in cases where there is no registration procedure for receiving a service, when the user information is recognized for the first time by the user information recognition unit 10, the user information may be stored as initial information.

[0060] The current emotion recognition unit 12 recognizes the current emotion, which is the emotion of the user U when he / she wishes to receive a service from the provision system S (i.e., when he / she wishes to obtain cooking information). The current emotion recognition unit 12 has an emotion model presentation unit 12a and a coordinate recognition unit 12b.

[0061] The emotion model presentation unit 12a presents to the user U, via the user terminal 2, an emotion model, which is a planar model whose coordinates are defined based on a plurality of basic emotions, in a format in which the user U can select any of the coordinates in the emotion model (see FIG. 8).

[0062] Here, the "basic emotions" could be, for example, joy, trust, fear, surprise, sadness, disgust, anger, and anticipation, which are used in Plutchik's wheel of emotions, or joy, fear, surprise, sadness, disgust, and anger, which are used in Ekman's theory.

[0063] The provision system S employs an emotion model including a first coordinate axis representing a plurality of basic emotions and a second coordinate axis representing the strength of the basic emotions. Specifically, the provision system S employs the EmojiGrid (registered trademark) shown in FIG. 3A as the emotion model.

[0064] It should be noted that the emotion model in the present invention is not limited to such an EmojiGrid, but may be any plane or three-dimensional model in which coordinates are defined based on a plurality of basic emotions.

[0065] Therefore, for example, as in the first modified example shown in Fig. 3B, Plutchik's Wheel of Emotions may be used as an emotion model. Furthermore, a model that uses the color arrangement of Plutchik's Wheel of Emotions but omits the letters may be used. Conversely, a model that uses the letter arrangement of Plutchik's Wheel of Emotions but omits the colors may be used.

[0066] Also, for example, a face scale may be adopted as the emotion model, as in the second modified example shown in Fig. 3C. Also, in the example shown in Fig. 3C, the emotion model is configured using only one type of face scale, but it may be configured by arranging multiple types of face scales.

[0067] Furthermore, in the present embodiment, the first variant, and the second variant, a planar model is used as the emotion model. However, the emotion model of the present invention is not limited to such a configuration, and may be a three-dimensional model.

[0068] However, the emotion model of the present invention is not limited to such a configuration. For example, even if a planar or three-dimensional model in which coordinates are defined based on a plurality of basic emotions is adopted, an emotion model that does not include coordinate axes may be adopted. Specifically, a model in which a plurality of colors are simply arranged at predetermined positions may be adopted as the emotion model.

[0069] Furthermore, the coordinate axes of the emotion model may be a combination of an axis indicating emotional valence intensity and an axis indicating arousal level.

[0070] Here, "valence" is a bipolar concept that indicates the qualitative difference in the emotion evoked (for example, positive valence such as attraction (pleasure) to the target object, or negative valence such as disgust (unpleasant)). Also, "valence intensity" is the strength of the emotion felt (Harabe, Yoshihiro (2015). Measuring the intensity of general valence concepts using paired comparison method: A preliminary study using unpleasant emotions. Tezukayama Gakuin University Faculty of Human Sciences Annual Research Report, (17)).

[0071] In addition, "arousal" here is a concept with two extremes, excitement (high arousal) and sedation (low arousal), and indicates the strength of the emotion evoked (Bradley, MM, Greenwald, MK, Petry, MC, & Lang, PJ (1992). Remembering Pictures: Pleasure and Arousal in Memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, 379-390.).

[0072] The coordinate recognition unit 12b recognizes selected coordinates, which are coordinates in an emotion model selected by the user U via the input unit 20 of the user terminal 2. In the providing system S, the coordinate recognition unit 12b recognizes, as selected coordinates, coordinates specified by the user U by touching the emotion model displayed on the touch panel, which is the input unit 20 and output unit 21 of the user terminal 2.

[0073] In addition, when adopting a face scale as in the second modified example shown in Fig. 3C, only the face image (button) constituting the face scale may be made selectable, and the coordinates of the selected face scale may be recognized as the selected coordinates. In addition, any coordinates within the range in which the face scale is displayed may be made selectable.

[0074] The coordinate recognition unit 12b then recognizes the recognized selected coordinates themselves as the current emotion, and therefore does not particularly identify what emotion is based on the coordinates.

[0075] The reason why the provision system S uses such a configuration for recognizing current emotions is as follows.

[0076] In other words, it is generally difficult for users to accurately express their own emotions (and thus to input them into a system). In addition, the user feels troubled when considering how to accurately express his or her own emotions, which may result in further changes in emotions.

[0077] Therefore, in the provision system S, an emotion model, which is a planar or three-dimensional model whose coordinates are defined based on a plurality of basic emotions, is adopted as an interface for inputting the emotions of the user U, and the coordinates selected in the emotion model are recognized as the emotions of the user U.

[0078] This allows the user U to easily and accurately express his / her own feelings as coordinate values ​​without converting them into linguistic expressions, etc. Furthermore, since the coordinate parameter is clearer than the emotion parameter, using the coordinate parameter as input data eliminates ambiguity from the input data.

[0079] Furthermore, in the provision system S, the emotion model employs a model including a first coordinate axis representing one or more basic emotions and a second coordinate axis representing the strength of the basic emotion.

[0080] This makes it easier for the user U to intuitively understand what emotion the coordinates in the emotion model indicate. In turn, it becomes easier for the user U to accurately express his / her own emotion as coordinates, further eliminating ambiguity from the input data.

[0081] The current emotion recognition unit of the present invention is not limited to such a configuration, and may be any device capable of recognizing the current emotion, which is the emotion at the time when the user desires to obtain cooking information. Therefore, as a method for recognizing the current emotion, for example, a questionnaire that simply asks about the current emotion may be used without using coordinate data.

[0082] The selection reason recognition unit 13 recognizes the selection reason from among a plurality of candidate reasons by using a predetermined first prediction model.

[0083] Here, the "candidate reason" is a reason that can be a selection reason. The "candidate reason" is a reason used to learn the first prediction model. The "selection reason" is a reason that influences the selection of a dish by the user U at the time when the current emotion is input, for example.

[0084] Potential reasons include, for example, "It makes me feel different from usual," "It's fun to enjoy with my family," "I can enjoy the cooking process," "I want to eat it myself," "My spouse's preference," "It goes well with alcohol," and "It suits the season / climate."

[0085] The first prediction model used by the selection reason recognition unit 13 uses each of the multiple candidate dishes, the user information recognized by the user information recognition unit 10, and the selection coordinates recognized by the current emotion recognition unit 12 as learning data, and machine-learns the correlation between the candidate reason, the user information, and the selection coordinates.

[0086] Here, "learning data" refers to data used to train a machine learning algorithm, and examples of such data include feature values, correct answer data, or combinations thereof used in the training. More specifically, "learning data" refers to data used to determine parameters held by a machine learning algorithm, and examples of such data include feature values, correct answer data, or combinations thereof used to determine the parameters.

[0087] When the user information recognized by the user information recognition unit 10 and the current emotion recognized by the current emotion recognition unit 12 are input to this first prediction model, a selection reason is extracted from multiple candidate reasons and output.

[0088] It should be noted that the selection reason recognition unit of the present invention is not limited to such a configuration, and may also recognize selection reasons, which are reasons that influence the user's choice of dish, using processing other than processing using the first prediction model.

[0089] Therefore, for example, instead of the first prediction model, the recommended dish may be recognized based on correlation data representing the correlation between the user information and each of the multiple candidate reasons. Note that, like the first prediction model, the correlation data preferably represents the correlation between the user information and the current emotion and each of the multiple candidate reasons. Furthermore, the data representing the current emotion may be coordinate-based as in the present embodiment.

[0090] Specifically, correlation data (see FIG. 4) showing the relationship between the current emotion (selected coordinates) and the scores for multiple candidate reasons may be prepared in advance for each user attribute (e.g., age, gender, place of origin, food allergies, etc.), and when recognizing the reason for selection, correlation data according to the attribute corresponding to the user information may be selected, the score may be recognized based on the current emotion (selected coordinates) by referring to the correlation data, and the reason for selection may be recognized from the candidate reasons listed in the correlation data based on the score.

[0091] In addition, for example, instead of the first predictive model, a generative model (e.g., a large-scale language model) may be used that uses first text data expressing reasons influencing the selection of a dish and user information as learning data, machine-learns the correlation between the first text data and the user information, and, when user information is input, generates and outputs second text data expressing reasons for selection that corresponds to the user information.

[0092] The generation model may be a model constructed by the service provider of the service that provides cooking information, or a model used by another person. The generation model is preferably a model that learns the correlation between the user information and the current emotion and each of the multiple candidate reasons, similar to the first prediction model. The data representing the current emotion may be coordinates as in the present embodiment.

[0093] When such a generative model is used, the reason for selection is not selected from a predetermined number of candidate reasons (i.e., fixed keywords) as in the above-mentioned first prediction model and correlation data in which a reason for selection is selected from a number of candidate reasons, but is text data that is generated each time the process is executed, so that the types of selection reasons provided can be more diverse than when such a first prediction model and correlation data are used.

[0094] The user information input to the generative model may be in the form of a list, such as "Favorite food preference: rich," or in the form used in a conversation, such as "I prefer rich foods over light foods."

[0095] In addition, the text data expressing the reasons may be in a bullet-point format, such as "Recommended dish: cream sauce, Reason: Because I like rich flavors and Western food," or in a conversational format, such as "Since I like rich flavors and prefer Japanese food to Western food, pasta with cream sauce would suit my taste. Cream sauce has a rich, rich flavor."

[0096] On the other hand, if a selection is made from a predetermined number of candidate reasons (i.e., fixed keywords), as in the case of the first prediction model and correlation data described above, the types of selection reasons are fixed, and the accuracy of the selection reasons provided can be improved compared to when such a generative model is used.

[0097] The recommended dish recognition unit 14 recognizes a plurality of recommended dishes from among the plurality of candidate dishes using a predetermined second prediction model.

[0098] Here, the "candidate reason" is a dish used in the second prediction model and is a dish that can become a recommended dish. Also, the "recommended dish" is a dish that is recommended for the user U to eat, as described above.

[0099] The second prediction model used by the recommended dish recognition unit 14 uses each of the multiple candidate dishes, the user information recognized by the user information recognition unit 10, the selection coordinates recognized by the current emotion recognition unit 12, and the reason for selection recognized by the selection reason recognition unit 13 as learning data, and machine-learns the correlation between the candidate dishes and the user information, the selection coordinates, and the reason for selection.

[0100] When the user information recognized by the user information recognition unit 10, the current emotion recognized by the current emotion recognition unit 12, and the reason for selection recognized by the reason for selection recognition unit 13 are input to this second prediction model, a recommended dish is extracted from among multiple candidate dishes and output.

[0101] In this way, by adopting a predictive model that uses not only user information and current emotions but also the reason for selection as input data as the second predictive model that outputs recommended dishes, the recommended dishes that are output can be not only suitable for each individual user U, but also dishes that are optimized for various situations (and therefore diverse).

[0102] In addition, the recommended dish recognition unit of the present invention may be any unit capable of recognizing recommended dishes, and is not necessarily limited to a configuration that recognizes recommended dishes using a prediction model such as the second prediction model of the provision system S.

[0103] Therefore, for example, instead of using a predictive model, the recommended dishes may be recognized based on correlation data that represents the correlation between the user information, current emotion, and selection reason, and each of a number of candidate dishes that may become recommended dishes. Also, the recommended dishes may be recognized without using the selection reason. However, in this case, a separate process is required to recognize the correlation between the recommended dishes and the selection reason.

[0104] The dish information storage unit 15 stores multiple pieces of information about the candidate dishes. The types of information correspond to the dish information described above. Specifically, the stored information includes information about the dish itself, such as the name, ingredients, and recipe of the dish, as well as information about the store where the dish is served and information about the means for obtaining the dish, such as the name of the service that can deliver the dish.

[0105] In the provision system S, the dish information storage unit 15, which is the dish information recognition unit, is configured as a part of the provision system S. However, the dish information recognition unit of the present invention is not limited to such a configuration, and may be configured to recognize multiple pieces of information related to candidate dishes.

[0106] Therefore, for example, the cooking information recognition unit may be provided in a system independent of the cooking information providing system. Furthermore, instead of a database that stores cooking information, the cooking information recognition unit may be a processing unit that searches information on the Internet, collects multiple pieces of cooking information, and recognizes them.

[0107] The dish information providing unit 16 obtains information about each of the multiple recommended dishes recognized by the recommended dish recognizing unit 14 from the dish information storage unit 15 as dish information to be presented to the user U. At the same time, the dish information providing unit 16 recognizes a selection reason corresponding to each of the multiple recommended dishes from among the selection reasons recognized by the selection reason recognizing unit 13.

[0108] Then, the dish information providing unit 16 provides the user U with dish information corresponding to each of the acquired multiple recommended dishes, including the selection reasons for each recommended dish, via the output unit 21 of the user terminal 2 (see Figures 9 and 10).

[0109] At this time, the cooking information providing unit 16 may change the way in which cooking information is provided depending on the current user information of the user U recognized by the user information recognizing unit 10. For example, if the user information includes information that the current eating behavior is related to cooking at home, the cooking information providing unit 16 provides a recipe for that dish as cooking information. On the other hand, if the user information includes information that the current eating behavior is related to eating out, the cooking information providing unit 16 does not provide a recipe for that dish as cooking information, but provides a restaurant that serves that dish.

[0110] [Processing performed in each processing section] Next, with reference to FIG. 2 and FIG. 5 to FIG. 10, a dish information providing method which is a process executed by the providing system S which is a computer system equipped with the above-mentioned processing unit will be described.

[0111] In this method of providing dish information, a process of recognizing the reason for selection is first performed, and then a process of providing dish information is performed.

[0112] In the process of recognizing the reason for selection, first, the user information recognition unit 10 presents a questionnaire for confirming current user information to the user U in a form that the user can answer, via the user terminal 2 (FIG. 5 / STEP 101).

[0113] Specifically, when a service provided by the provision system S is used for the first time, a questionnaire asking for information for identifying the user U (for example, a separately registered account name) is first presented in a form that the user can answer.

[0114] Once the questionnaire has been answered, or when the user uses the service for the second or subsequent time, a questionnaire is presented in an answerable format, as shown in Figure 7, asking about the user U's age (biometric information), number of people living with them, whether they cook at home or eat out (eating environment), and favorite ingredients (preferences).

[0115] Next, when the questionnaire is answered, the user information recognition unit 10 recognizes the user information of the current user U based on the questionnaire answer and the existing user information (FIG. 5 / STEP 102).

[0116] Next, the user information recognition unit 10 judges whether the current user information differs from the known user information (FIG. 5 / STEP 103).

[0117] If the current user information differs from the known user information (YES in STEP 103), the user information recognition unit 10 updates the user information by storing the current user information in the user information storage unit 11 (FIG. 5 / STEP 104).

[0118] Specifically, at the time of first use, the user information recognition unit 10 recognizes the current user information of the user U based on information obtained separately at the time of account registration, etc., and the answers to the current questionnaire. At this time, the user information has not yet been stored, and the current user information is naturally different from the known user information, so the user information recognition unit 10 stores the current user information as it is in the user information storage unit 11 as the user information of the user U.

[0119] Furthermore, when using the service for the second or subsequent times, the user information recognition unit 10 first compares the current user information of the recognized user U with the user information of the user U from the previous time or earlier acquired from the user information storage unit 11. If there is a difference between the user information, the user information recognition unit 10 creates new user information that adopts the current user information for the difference, and stores the new user information in the user information storage unit 11 as the user information of the user U.

[0120] If the current user information does not differ from the previously known user information (NO in STEP 103), or if new user information is stored in the user information storage unit 11 (when STEP 104 is executed), the current emotion recognition unit 12 presents an emotion model and coordinates in the emotion model in a selectable format to the user U via the user terminal 2 (FIG. 5 / STEP 105).

[0121] Specifically, as shown in Fig. 8, the current emotion recognition unit 12 displays an emotion model (EmojiGrid) on the touch panel of the user terminal 2. When the user touches any point on this emotion model, the current emotion recognition unit 12 determines that the coordinates of that point have been selected.

[0122] Next, the current emotion recognition unit 12 judges whether any coordinates in the emotion model have been selected (FIG. 5 / STEP 106).

[0123] If no coordinates have been selected (NO in STEP 106), the current emotion recognition unit 12 repeats the determination in STEP 106 at a predetermined control period.

[0124] On the other hand, if any coordinates are selected (YES in STEP 106), the current emotion recognition unit 12 recognizes the coordinates as selected coordinates, and recognizes the selected coordinates as the current emotion (FIG. 5 / STEP 107).

[0125] Next, the reason for selection recognition unit 13 inputs the current user information and the current emotion to the first prediction model, and outputs the reason for selection for the user U (FIG. 5 / STEP 108), and ends the current process.

[0126] In the process of providing dish information, first, the user information recognition unit 10 recognizes the user information and the current emotion recognition unit 12 recognizes the current emotion by a process similar to that for recognizing the reason for selection (FIG. 6 / STEP 201).

[0127] Next, the recommended dish recognition unit 14 inputs the current user information and current emotion, as well as the reason for selection recognized by the reason for selection recognition unit 13, into the second prediction model, and outputs a number of recommended dishes for the user U (Figure 6 / STEP 202).

[0128] Next, the cooking information provider 16 acquires cooking information for each of the plurality of recommended dishes from the cooking information storage unit 15 (FIG. 6 / STEP 203).

[0129] Next, as shown in FIG. 9, the dish information provider 16 provides the user U with the dish name, which is one of the acquired dish information, in a selectable format via the user terminal 2 (FIG. 6 / STEP 204).

[0130] Next, the dish information providing unit 16 determines whether any dish name has been selected (FIG. 6 / STEP 204).

[0131] If no dish name has been selected (NO in STEP 204), the dish information provider 16 repeats the determination in STEP 204 at a predetermined control cycle.

[0132] On the other hand, if any of the dish names is selected (YES in STEP 204), the dish information providing unit 16 provides detailed dish information about the recommended dish corresponding to the selected dish name to the user U via the user terminal 2 (Figure 6 / STEP 205), and then ends this processing.

[0133] 10, for example, if the current user information includes information that the eating behavior is related to home cooking, the cooking information providing unit 16 displays a recipe for a recommended dish corresponding to the selected dish name on the touch panel of the user terminal 2. If the information includes information that the eating behavior is related to eating out, the cooking information providing unit 16 displays a store or service that offers a recommended dish corresponding to the selected dish name on the touch panel.

[0134] Furthermore, in addition to the recipe, store, etc., the dish information also displays a selection reason corresponding to the recommended dish as a reason why the recommended dish is recommended. In this embodiment, as shown in Fig. 10, a sentence is first generated based on the selection reasons "It's easy because you seem busy" and "Kids will like it", and then the selection reason is presented in the form of presenting the sentence.

[0135] As described above, the dish information providing method executed by the provision system S provides dish information regarding recommended dishes to the user U in a form that includes the selection reasons, which are reasons that influence the selection of dishes by the user U, that correspond to the recommended dishes. In other words, the selection reasons are provided as grounds for recommending the recommended dishes.

[0136] Here, the selection reason is output by converting user information (e.g., environmental conditions of user U) into a form of reason using a first prediction model that performs machine learning on the correlation between each of a plurality of candidate reasons that are possible reasons for selection and the user information. This allows user U to directly grasp the reason for the recommendation of the recommended dish, compared to a case where user information is provided as is.

[0137] Therefore, according to this method of providing cooking information, the reasons for selection suitable for each individual user U are directly provided in the cooking information regarding the recommended dishes, so that the cooking information regarding the recommended dishes can be provided to the user U in a format that gives a sufficient sense of convincingness.

[0138] In addition, in the provision system S, in addition to user information such as the environment of the user U, the current emotion of the user U is also used as input data used in the first prediction model that outputs the reason for selection.

[0139] Generally, even in the same environment, if the user's emotions are different, the impact of the environment on the user will be different. It is also known that there is a certain correlation between emotions and eating behavior (see, for example, Imada Sumio (2009). Emotions and Eating Behavior, Emotion Psychology Research 17, 120-128). In other words, the selection reasons output by the first prediction model in the provision system S of the present invention are suitable for each individual user U.

[0140] However, the cooking status providing method of the present invention is not limited to such a configuration, and it is sufficient if the reason for selection, which is output by inputting user information, is included in the cooking information and provided. Therefore, it is not necessary to use the current emotion as input data. If the current emotion is not used as input data, the current emotion recognition unit may be omitted from the cooking status providing system.

[0141] [Other embodiments] Although the illustrated embodiment has been described above, the present invention is not limited to this embodiment.

[0142] For example, in the above embodiment, the selection reason is included in the dish information, which is information about the recommended dish, and is provided to the user U. However, the selection reason may be provided to the user independently, separate from the dish information.

[0143] In the above embodiment, the provision system S is a single computer system. However, the present invention also includes a dish information provision program for executing the above-mentioned dish information provision method by any one or more computers, and a recording medium on which the program is recorded and which is readable by a computer used by a user or the like.

[0144] Furthermore, although not adopted in the above embodiment, the selection situation, which is the situation at the time when the dish information (and, by extension, the selection reason and corresponding recommended dish contained therein) is selected, may be recognized, and each time the selection situation is recognized, or when the selection situation has been recognized a predetermined number of times, the first prediction model, the second prediction model, or the generation model may be re-learned, or the correlation data may be updated, using the selection situation recognized up to that point.

[0145] Here, the "selection status" includes the number of times the dish information was selected, the display position of the selected dish information, as well as dish information selected before the dish information was selected and other dish information that was presented at the same time as the dish information was selected.

[0146] Other examples of "selection status" include the time when the dish information was selected (and therefore whether it was breakfast or dinner, etc.), the user's actions before and after the dish information was selected (e.g., the contents of operations on the mobile terminal, etc.) Furthermore, the contents of operations on the mobile terminal include not only those that the user consciously performed on the terminal (e.g., operations on the touch panel, etc.), but also those that the user unconsciously performed (e.g., the time spent viewing dish information, etc.).

[0147] Furthermore, the selection status used for re-learning the first prediction model, the second prediction model, or the generative model, or for updating correlation data does not necessarily need to use all previous selection statuses, and only the selection status at a specified timing (for example, the most recent specified number of times, within a specified period, etc.) may be used.

[0148] In this way, if the selection situation, which is the situation at the time when the dish information (and, by extension, the selection reason and corresponding recommended dish contained therein) was selected, is used to re-learn the first prediction model, the second prediction model, or the generation model, or to update the correlation data, the first prediction model, the second prediction model, the generation model, or the correlation data after the re-learning or update will be more in line with the user's actual situation than before the re-learning or update.

[0149] As a result, the cooking information output using the first prediction model, the second prediction model, the generative model, or the correlation data after the re-learning or update will be more suitable for the user than the cooking information output using the ones before the re-learning or update.

[0150] In addition, being selected as cooking information means that it is highly likely that the user directly likes it. In addition, the more times the information is selected, the higher the possibility that the information is directly liked. Therefore, by using the number of selections as the selection status, the cooking information output using the first prediction model, the second prediction model, the generative model, or the correlation data that has been re-learned or updated using the selection status becomes more suitable for the user. [Explanation of symbols]

[0151] 1...server, 2...user terminal, 10...user information recognition unit, 11...user information storage unit, 12...current emotion recognition unit, 12a...emotion model presentation unit, 12b...coordinate recognition unit, 13...selection reason recognition unit, 14...recommended dish recognition unit, 15...dish information storage unit (dish information recognition unit), 16...dish information provision unit, 20...input unit, 21...output unit, S...provision system (dish information provision system), U...user.

Claims

1. A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit inputting the user information and outputting selection reasons, which are reasons that influence the user's selection of a dish, using a first prediction model; and a recommended dish recognition unit recognizing the recommended dishes; a cooking information providing unit providing the user with the cooking information including the selection reason corresponding to the recommended cooking dish; A method for providing cooking information, characterized in that the first prediction model uses each of a plurality of candidate reasons that are reasons that may be the reason for selection and the user information as learning data, performs machine learning to determine the correlation between the candidate reasons and the user information, and selects and outputs the reason for selection from the plurality of candidate reasons.

2. The method for providing food information according to claim 1, a current emotion recognition unit for recognizing a current emotion that is an emotion at a time when the user desires to acquire the cooking information; The cooking information providing method is characterized in that the first prediction model uses each of the multiple candidate reasons, the user information, and the current emotion as learning data, performs machine learning to detect correlations between the candidate reasons and the user information and the current emotion, and selects and outputs the reason for selection from the multiple candidate reasons.

3. The method for providing food information according to claim 2, the current emotion recognition unit has an emotion model presenting unit that presents to the user an emotion model, which is a planar or three-dimensional model whose coordinates are defined based on a plurality of basic emotions, in a form in which any coordinate in the emotion model can be selected, and a coordinate recognition unit that recognizes selected coordinates, which are coordinates in the emotion model selected by the user, as the current emotion; A method for providing cooking information, characterized in that the first prediction model uses each of the multiple candidate reasons, the user information, and the selected coordinates as learning data, performs machine learning to determine the correlation between the candidate reasons, the user information, and the selected coordinates, and selects and outputs the selection reason from the multiple candidate reasons.

4. The method for providing food information according to claim 3, A method for providing cooking information, characterized in that the emotion model is a model including a first coordinate axis representing one or more basic emotions, and a second coordinate axis representing the strength of the basic emotions.

5. The method for providing food information according to claim 1, the recommended dish recognition unit recognizes the recommended dish using a second prediction model that inputs the user information and the reason for selection and outputs the recommended dish; The second prediction model uses each of a plurality of candidate dishes that may become the recommended dish, as well as the user information and the reason for selection as learning data, and performs machine learning to detect correlations between the candidate dishes and the user information and the reason for selection, and selects and outputs the recommended dish from the plurality of candidate dishes.

6. A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit recognizing a selection reason, which is a reason that influences the user's selection of a dish, based on correlation data representing a correlation between the user information and each of a plurality of candidate reasons; a recommended dish recognition unit recognizing the recommended dishes; A method for providing cooking information, comprising a step in which a cooking information providing unit provides the user with cooking information including the selection reason corresponding to the recommended dish.

7. A method for providing a user with recommended dishes, the recommended dishes being dishes that are recommended for the user to eat, comprising: A step of recognizing user information including at least one of biometric information, preferences, and an environment in the eating behavior of the user by a user information recognition unit; a selection reason recognition unit inputting the user information and outputting selection reasons, which are reasons influencing the user's selection of a dish, using a generative model to output the selection reasons; a recommended dish recognition unit recognizing the recommended dishes; a cooking information providing unit providing the user with the cooking information including the selection reason corresponding to the recommended cooking dish; The method for providing cooking information is characterized in that the generative model uses first text data expressing reasons influencing the selection of a dish and the user information as learning data, performs machine learning to determine the correlation between the first text data and the user information, and generates and outputs second text data expressing the reasons for selection.

8. A cooking information providing system that provides a user with cooking information that is information about recommended cooking, which is cooking that is recommended for a user to eat, comprising: The system includes a user information recognition unit, a selection reason recognition unit, a recommended dish recognition unit, and a dish information provision unit, A cooking information providing system configured to execute the cooking information providing method according to any one of claims 1 to 7.

9. A cooking information providing device that provides a user with cooking information that is information about recommended cooking, which is cooking that is recommended for a user to eat, comprising: The system includes a user information recognition unit, a selection reason recognition unit, a recommended dish recognition unit, and a dish information provision unit, A cooking information providing device configured to execute the cooking information providing method according to any one of claims 1 to 7.

10. A recipe information providing method for providing a user with recipe information, the recipe information being information about recommended recipes that are recommended for a user's eating behavior, comprising: A cooking information providing program that causes the computer to execute the cooking information providing method according to any one of claims 1 to 7.

11. 11. A recording medium having the cooking information providing program according to claim 10 recorded thereon, said recording medium being readable by said computer.

Images