Information processing apparatus, method for producing brewed beverage, learning model generation method, information output method, and information display method

Abstract

To provide an information processing device, a brewed beverage manufacturing method, a learning model generation method, an information output method, and an information display method capable of efficiently and stably improving flavor of a brewed beverage.SOLUTION: A first model generation unit configured to generate a first learning model by machine learning using a plurality of first training data including sensory evaluation data of a brewed beverage and comprehensive evaluation data indicating comprehensive evaluation of a flavor of the brewed beverage with respect to a predetermined flavor, and a second model generation unit configured to generate a second learning model by machine learning using a plurality of second training data including process data related to a brewing process of the brewed beverage and analysis value data of the brewed beverage, and the sensory evaluation data.SELECTED DRAWING: Figure 11

Description

[Technical Field]

[0001] The present disclosure relates to an information processing device, a method for producing a brewed beverage, a learning model generation method, an information output method, and an information display method. [Background technology]

[0002] Conventionally, many beverages (hereinafter simply referred to as beverages or brewed beverages) have been produced through fermentation during the production of beer, wine, sake, etc. It is known that various aroma components are involved in the flavor and aroma of such brewed beverages (see Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 5894027 Summary of the Invention [Problem to be solved by the invention]

[0004] Here, the brewing process for the above-mentioned brewed beverage involves, for example, complex chemical or biological factors, making it difficult to grasp the correlation between each flavor and each brewing process in the brewed beverage. Therefore, maintaining or improving the flavor of the above-mentioned brewed beverage may be performed based on, for example, the knowledge and empirical rules of a brewing technician, which may be insufficient in terms of efficiency and stability.

[0005] Therefore, one disclosure provides an information processing device, a method for producing a brewed beverage, a learning model generation method, an information output method, and an information display method that enable the efficient and stable maintenance and improvement of the flavor of brewed beverages. [Means for solving the problem]

[0006] The information processing device of the present disclosure has a first model generation unit that generates a first learning model by machine learning using a plurality of first teacher data including sensory evaluation data of a brewed beverage and overall evaluation data indicating an overall evaluation of the flavor of the brewed beverage relative to a specified flavor, and a second model generation unit that generates a second learning model by machine learning using a plurality of second teacher data including either process data related to the brewing process of the brewed beverage or analytical value data of the brewed beverage, and the sensory evaluation data. [Effects of the Invention]

[0007] The information processing device, brewed beverage manufacturing method, learning model generation method, information output method, and information display method disclosed herein make it possible to efficiently and stably maintain and improve the flavor of brewed beverages. [Brief explanation of the drawings]

[0008] [Figure 1] FIG. 1 is a diagram illustrating a configuration of an information processing system 100 according to the first embodiment. [Figure 2] FIG. 2 is a diagram illustrating the hardware configuration of the information processing device 1 according to the first embodiment. [Figure 3] FIG. 3 is a block diagram of the functions of the information processing device 1 according to the first embodiment. [Figure 4] FIG. 4 is a diagram illustrating the relationship between the first teacher data DT11 and the first learning model MD1. [Figure 5] FIG. 5 is a diagram illustrating the relationship between the second teacher data DT12 and the second learning model MD2. [Figure 6] FIG. 6 is a diagram illustrating the relationship between the third teacher data DT13 and the third learning model MD3. [Figure 7] FIG. 7 is a flowchart illustrating the model generation process according to the first embodiment. [Figure 8] FIG. 8 is a diagram illustrating the model generation process according to the first embodiment. [Figure 9]FIG. 9 is a diagram illustrating the model generation process according to the first embodiment. [Figure 10] FIG. 10 is a diagram illustrating the model generation process according to the first embodiment. [Figure 11] FIG. 11 is a flowchart illustrating the information output process according to the first embodiment. [Figure 12] FIG. 12 is a diagram illustrating the information output process according to the first embodiment. [Figure 13] FIG. 13 is a diagram illustrating the information output process according to the first embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0009] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. However, such descriptions should not be interpreted in a limiting sense, and do not limit the subject matter described in the claims. Furthermore, various changes, substitutions, and modifications can be made without departing from the spirit and scope of the present disclosure. Furthermore, different embodiments can be combined as appropriate.

[0010] [Information Processing System 100 in the First Embodiment] First, a description will be given of an example of the configuration of an information processing system 100 according to the first embodiment. Fig. 1 is a diagram illustrating the configuration of the information processing system 100 according to the first embodiment.

[0011] As shown in FIG. 1, the information processing system 100 includes, for example, an information processing device 1 and an operation terminal 5.

[0012] The information processing device 1 is, for example, a physical machine or a virtual machine. The operation terminal 5 is, for example, a mobile terminal such as a PC (Personal Computer) or a smartphone, and is a terminal through which a brewing technician inputs necessary information. The information processing device 1 and operation terminal 5 can access each other via a network NW such as the Internet.

[0013] The information processing device 1 generates a learning model (hereinafter also referred to as the first learning model) that can output (predict) overall evaluation data as sensory evaluation data is input, by machine learning using multiple training data (hereinafter also referred to as the first training data), each of which includes sensory evaluation data (hereinafter also referred to as the sensory evaluation data) indicating a sensory evaluation of the beverage to be processed (hereinafter also referred to as the target beverage) and overall evaluation data (hereinafter also referred to as the overall evaluation data) indicating an overall evaluation of the flavor of the target beverage against predetermined indicators (hereinafter also referred to as the overall evaluation).

[0014] Here, the overall evaluation is, for example, a comprehensive evaluation of the flavor index to be evaluated, and may be, for example, an evaluation of the degree of preference or the degree of difference as described in the BCOJ Sensory Evaluation Method (published by the Brewery Society of Japan, edited by the International Technical Committee of the Brewers Association of Japan, 2002). The overall evaluation data may be, for example, data indicating a value calculated from the total score of the sensory evaluation data, or data indicating a value that quantifies the degree of preference in the user's evaluation (for example, data indicating a score when the user rates how much they like the product to be evaluated on a 10-point scale).

[0015] Specifically, the first training data is training data including, for example, one or more types of sensory evaluation data and one overall evaluation data. The first learning model is, for example, a learning model that can output (predict) one overall evaluation data, which is a response variable, in response to input of one or more types of sensory evaluation data, which is an explanatory variable.

[0016] The first learning model is a learning model that can output, for example, the contribution (importance) of each explanatory variable to a target variable. Specifically, for example, a decision tree, a random forest model, XGBoost (eXtreme Gradient Boosting), LightGBM (Light Gradient Boosting Machine), GBR, etc. can be applied as the first learning model.

[0017] Furthermore, the information processing device 1 generates a learning model (hereinafter also referred to as a second learning model) that can output (predict) sensory evaluation data in response to input of at least one of process data (hereinafter also referred to as simply process data) related to the brewing process of the target beverage and analytical value data (hereinafter also referred to as simply analytical value data) through machine learning using a plurality of training data (hereinafter also referred to as second training data), each of which includes sensory evaluation data and at least one of process data and analytical value data indicating analytical values ​​for the target beverage. The brewing process is, for example, at least one of a plurality of steps related to the production process of a brewed beverage, such as a preparation process, a fermentation process, and a process for adjusting the ingredient combination.

[0018] Specifically, the second training data is training data including, for example, at least one of one or more types of process data and one or more types of analytical value data, and one or more types of sensory evaluation data. The second learning model is, for example, a learning model capable of outputting (predicting) one or more types of sensory evaluation data, which is a response variable, in response to input of at least one of one or more types of process data and one or more types of analytical value data, which are explanatory variables.

[0019] The second learning model is, for example, a learning model that can output the contribution (importance) of each explanatory variable to the objective variable, similar to the first learning model. Specifically, for example, a decision tree, a random forest model, XGBoost, LightGBM, GBR, etc. can be applied to the second learning model.

[0020] Furthermore, the information processing device 1 identifies, for example, a type (hereinafter also referred to as the first type) of sensory evaluation data included in the first teacher data whose contribution to the overall evaluation data in the first learning model satisfies a predetermined condition (hereinafter also referred to as the first condition).Then, the information processing device 1 identifies, for example, a type (hereinafter also referred to as the second type) of at least one type of process data and analysis value data included in the second teacher data whose contribution to the sensory evaluation data of the first type in the second learning model satisfies a predetermined condition (hereinafter also referred to as the second condition).After that, the information processing device 1 outputs, for example, information indicating the identified second type.

[0021] That is, the information processing device 1 in this embodiment, for example, identifies a type (first type) of sensory evaluation data that has a high contribution to overall evaluation data in a first learning model for the target beverage as a type of sensory evaluation data that is important in bringing the flavor of the target beverage closer to a predetermined flavor (in other words, a target flavor).The information processing device 1 in this embodiment then outputs a type (second type) of process data or analytical value data that has a high contribution to the first type of sensory evaluation data in a second learning model for the target beverage as a type of process data or analytical value data that is important in adjusting the first type of sensory evaluation data.

[0022] This enables the information processing device 1 in this embodiment to output, as result information, for example, the type of process data or the type of analytical value data that is important when bringing the flavor of the target beverage closer to a predetermined flavor.

[0023] Therefore, brewing engineers can efficiently and stably make adjustments to maintain or improve the flavor of the target beverage by, for example, referring to the information (type of process data and type of analytical value data) output by the information processing device 1. Specifically, brewing engineers can efficiently review the brewing process to maintain or improve the flavor of the target beverage.

[0024] The subject beverages in this disclosure are, for example, beverages produced by fermentation, i.e., brewed beverages. Non-limiting examples of brewed beverages include beer-flavored beverages (e.g., beer and happoshu), wine, sake, and other brewed alcoholic beverages, as well as RTD (Ready-to-Drink) beverages made from brewed alcoholic beverages.

[0025] [Information processing device 1 according to the first embodiment] Next, a description will be given of the configuration of the information processing device 1 in the first embodiment. Fig. 2 is a diagram illustrating the hardware configuration of the information processing device 1 in the first embodiment.

[0026] 2, the information processing device 1 is a computer device having, for example, a CPU 101 which is a processor, a memory 102, a communication device 103, and a storage medium 104. Each unit is connected to each other via, for example, a bus 105.

[0027] The storage medium 104 has a program storage area (not shown) that stores a program 110 for, for example, a process for generating each learning model (hereinafter also referred to as a model generation process) and a process for outputting types of process data and types of analytical value data that can efficiently adjust the flavor of a target beverage (hereinafter also referred to as an information output process). Hereinafter, the model generation process and the information output process will be collectively referred to simply as the model generation process, etc. The storage medium 104 also has an information storage area 130 that stores various pieces of information used when performing the model generation process, etc. The storage medium 104 may be, for example, a hard disk drive (HDD) or a solid state drive (SSD).

[0028] The CPU 101 performs model generation processing and the like by executing a program 110 loaded from the storage medium 104 into the memory 102, for example.

[0029] The communication device 103 accesses, for example, the operation terminal 5 via the network NW.

[0030] The information processing device 1 may include, for example, a field programmable gate array (FPGA) or an application specific integrated circuit (ASIC), and the model generation process and the like may be executed in, for example, the FPGA or the ASIC.

[0031] Furthermore, the following description will be given assuming that the information processing system 100 has one information processing device 1, but the present invention is not limited to this. Specifically, the information processing system 100 may have, for example, a plurality of information processing devices 1. The model generation process and the like may be performed in a distributed manner among, for example, a plurality of information processing devices 1. Specifically, the model generation process and the information output process may each be executed by, for example, different information processing devices 1.

[0032] In addition, the following description will be given on the assumption that the model control process and the like are executed in the information processing device 1, but the present invention is not limited to this. Specifically, the model control process and the like may be executed in the operation terminal 5, for example.

[0033] [Functions of information processing device 1] Next, the functions of the information processing device 1 will be described. Fig. 3 is a block diagram of the functions of the information processing device 1 in the first embodiment. Fig. 4 is a diagram explaining the relationship between the first teacher data DT11 and the first learning model MD1. Fig. 5 is a diagram explaining the relationship between the second teacher data DT12 and the second learning model MD2. Fig. 6 is a diagram explaining the relationship between the third teacher data DT13 and the third learning model MD3.

[0034] As shown in FIG. 3, the information processing device 1 realizes various functions including an information management unit 111, a data generation unit 112, a model generation unit 113, a type identification unit 114, and an information output unit 115 by organically cooperating with hardware such as a CPU 101 and a memory 102 and programs.

[0035] In addition, as shown in Figure 3, the information processing device 1 stores, for example, sensory evaluation data DT1, process data DT2, analytical value data DT3, overall evaluation data DT4, a first learning model MD1, a second learning model MD2, and a third learning model MD3 in the information storage area 130.

[0036] The information management unit 111 stores in the information storage area 130, for example, the sensory evaluation data DT1, process data DT2, analysis value data DT3, and overall evaluation data DT4 input by the brewing engineer via the operation terminal 5.

[0037] 4, the data generation unit 112 generates a plurality of first teacher data DT11 each including, for example, sensory evaluation data DT1 and overall evaluation data DT4. Then, the information management unit 111 stores, for example, the plurality of first teacher data DT11 generated by the data generation unit 112 in the information storage area 130 (hereinafter also referred to as information storage area 130a).

[0038] 5, the data generating unit 112 generates a plurality of second teacher data DT12 each including, for example, sensory evaluation data DT1, process data DT2, and analysis value data DT3. Then, the information managing unit 111 stores, for example, the plurality of second teacher data DT12 generated by the data generating unit 112 in the information storage area 130a.

[0039] 6, the data generating unit 112 generates a plurality of third teacher data DT13 each including, for example, process data DT2 and analysis value data DT3. Specifically, the data generating unit 112 generates a plurality of third teacher data DT13 each including, for example, one or more types of process data DT2 and one or more types of analysis value data DT3. Then, the information managing unit 111 stores, for example, the plurality of third teacher data DT13 generated by the data generating unit 112 in the information storage area 130a.

[0040] In addition, the information management unit 111 may be configured to store at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 in the information storage area 130a, for example, when a brewing technician directly inputs at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 via the operation terminal 5.

[0041] Furthermore, the data generation unit 112 may, for example, accept data input by a brewing engineer and automatically generate at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 from the accepted input data. In this case, the information management unit 111 may, for example, store at least one of the generated first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 in the information storage area 130a.

[0042] The data generation unit 112 may also be configured to automatically collect data from, for example, manufacturing equipment installed in a factory or a database that aggregates information from the manufacturing equipment, and automatically generate at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 from the collected data. In this case, the information management unit 111 may be configured to store at least one of the generated first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 in the information storage area 130a.

[0043] 4, the model generation unit 113 generates a first learning model MD1 by machine learning of a plurality of first teacher data DT11 stored in the information storage area 130a. The first learning model MD1 is a learning model that can output (predict) one overall evaluation data DT4, which is a target variable, in response to input of one or more types of sensory evaluation data DT1, which are explanatory variables. The information management unit 111 then stores the first learning model MD1 generated by the model generation unit 113 in the information storage area 130 (hereinafter also referred to as the information storage area 130b).

[0044] 5, the model generation unit 113 generates a second learning model MD2 by machine learning of multiple pieces of second teacher data DT12 stored in the information storage area 130a. The second learning model MD2 is a learning model that can output (predict) one or more types of sensory evaluation data DT1, which is a response variable, in response to input of at least one of one or more types of process data DT2 and one or more types of analysis value data DT3, which are explanatory variables. The information management unit 111 then stores the second learning model MD2 generated by the model generation unit 113 in the information storage area 130b.

[0045] 6, the model generation unit 113 generates a third learning model MD3 by machine learning of multiple third teacher data DT13 stored in the information storage area 130a. The third learning model MD3 is a learning model that can output (predict) one or more types of analysis value data DT3, which are objective variables, in response to input of one or more types of process data DT2, which are explanatory variables. The information management unit 111 then stores the third learning model MD3 generated by the model generation unit 113 in the information storage area 130b.

[0046] The third learning model MD3 is a learning model that can output the contribution (importance) of each explanatory variable to the objective variable, similar to the first learning model MD1 and the second learning model MD2. Specifically, the third learning model can be, for example, a decision tree, a random forest model, XGBoost, LightGBM, or GBR.

[0047] Furthermore, the model generation unit 113 may, for example, divide the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 stored in the information storage area 130 into training data and test data, respectively, and use them.

[0048] The type identification unit 114, for example, identifies a first type from among the types of sensory evaluation data DT1 included in the first teacher data DT11 stored in the information storage area 130a, whose contribution to the overall evaluation data in the first learning model MD1 satisfies a first condition.

[0049] Specifically, the type identification unit 114 identifies, for example, as the first type, a type of sensory evaluation data DT1 included in the first teacher data DT11 whose contribution to the overall evaluation data in the first learning model MD1 is greater than or equal to a predetermined threshold (hereinafter also referred to as the first threshold).

[0050] In addition, the type identification unit 114 identifies, for example, a predetermined number (hereinafter also referred to as the first predetermined number) of types of sensory evaluation data DT1 included in the first teacher data DT11 that have the highest contribution to the overall evaluation data in the first learning model MD1 as the first type.

[0051] The first predetermined number may be, for example, the number of all types for the sensory evaluation data DT1 included in the first teacher data DT11. That is, the type identification unit 114 may be configured to identify, for example, all types for the sensory evaluation data DT1 included in the first teacher data DT11.

[0052] Then, the type identification unit 114 identifies a second type, for example, from at least one of the types of process data DT2 and analysis value data DT3 included in the second teacher data DT12 stored in the information storage area 130a, whose contribution to the sensory evaluation data DT1 of the first type in the second learning model MD2 satisfies the second condition.

[0053] Specifically, the type identification unit 114 identifies as the second type, for example, at least one of the types of process data DT2 and analysis value data DT3 included in the second teacher data DT12, the type whose contribution to the first type of sensory evaluation data DT1 in the second learning model MD2 is greater than or equal to a predetermined threshold value (hereinafter also referred to as the second threshold value).

[0054] In addition, the type identification unit 114 identifies as the second type, for example, a predetermined number (hereinafter also referred to as the second predetermined number) of types from at least one of the process data DT2 and analysis value data DT3 included in the second teacher data DT12, which types have the highest contribution to the sensory evaluation data DT1 of the first type in the second learning model MD2.

[0055] The second predetermined number may be, for example, the sum of the number of all types for the process data DT2 included in the second teacher data DT12 and the number of all types for the analysis value data DT3 included in the second teacher data DT12. That is, the type identification unit 114 may identify, for example, all types for the process data DT2 included in the second teacher data DT12 and all types for the analysis value data DT3 included in the second teacher data DT12.

[0056] The information output unit 115 outputs, for example, information (hereinafter also referred to as result information) indicating each of the second types identified by the type identification unit 114. Specifically, the information output unit 115 outputs, for example, information indicating the second type identified by the type identification unit 114 to the operation terminal 5.

[0057] In addition, the information output unit 115 may output, for example, information indicating the first type identified by the type identification unit 114 in addition to or instead of the information indicating the second type identified by the type identification unit 114.

[0058] [Model Generation Process in the First Embodiment] Next, the model generation process in the first embodiment will be described. Fig. 7 is a flowchart illustrating the model generation process in the first embodiment. Fig. 8 to Fig. 10 are diagrams illustrating the model generation process in the first embodiment.

[0059] As shown in FIG. 7, the model generation unit 113 waits, for example, until it is time to generate a model (NO in S11). The model generation timing may be, for example, the timing when a brewing engineer inputs information indicating that a model generation process will be performed. Alternatively, the model generation timing may be, for example, the timing when at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 generated by the information processing device 1 using data input by the brewing engineer is stored in the information storage area 130. Alternatively, the model generation timing may be, for example, the timing when at least one of the first teacher data DT11, the second teacher data DT12, and the third teacher data DT13 generated by the information processing device 1 using data collected from manufacturing equipment installed in a factory or a database that aggregates information from the manufacturing equipment is stored in the information storage area 130.

[0060] Then, when it is time to generate a model (YES in S11), the model generation unit 113 generates a first learning model MD1 (S12) by machine learning of the first teacher data DT11 (e.g., the first teacher data DT11 generated in advance by the data generation unit 112) stored in the information storage area 130. A specific example of the first teacher data DT11 will be described below.

[0061] [Specific example of the first teacher data DT11] FIG. 8 is a diagram illustrating a specific example of the first teacher data DT11.

[0062] The first teacher data DT11 shown in Figure 8 has, for example, items such as "overall evaluation" in which the overall evaluation data DT4 is set, and "burnt smell," "sour taste," and "fruity" in which the types of the sensory evaluation data DT1 are set.

[0063] Specifically, the first teacher data DT11 in the first row shown in Figure 8 has, for example, a setting of "7.0" for "overall evaluation," "3.0" for "burnt smell," "4.0" for "sour taste," and "5.0" for "fruity."

[0064] 8, for example, "5.0" is set as "overall evaluation," "2.0" is set as "burnt odor," "4.0" is set as "sour odor," and "3.0" is set as "fruity." Explanation of other data included in FIG. 8 will be omitted.

[0065] That is, in this case, the model generation unit 113 generates a first learning model MD1 that can output (predict) a value corresponding to the "overall evaluation" in response to input of values ​​corresponding to each of, for example, "burnt smell," "sour smell," and "fruity."

[0066] Returning to FIG. 7, the model generation unit 113 generates a second learning model MD2 by machine learning of the second teacher data DT12 (e.g., the second teacher data DT12 generated in advance by the data generation unit 112) stored in the information storage area 130 (S13). Note that S13 may be performed in parallel with S12, for example. Specific examples of the second teacher data DT12 will be described below.

[0067] [Specific example of the second teacher data DT12] FIG. 9 is a diagram illustrating a specific example of the second teacher data DT12.

[0068] The second teacher data DT12 shown in Figure 9 has, for example, items such as "burnt smell," "sour smell," and "fruity," which are set as types of sensory evaluation data DT1, "sugar content (degrees)," "pH," etc., which are set as types of analytical value data DT3, and "brewing kettle temperature (℃)" and "number of fermentation days (days)," etc., which are set as types of process data DT2.

[0069] Specifically, the second teacher data DT12 in the first row shown in Figure 9 has, for example, "3.0" set as "burnt smell," "4.0" set as "sour taste," "5.0" set as "fruity," "11.0 (degrees)" set as "sugar content (degrees)," "4.0" set as "pH," "62 (degrees)" set as "brewing kettle temperature (℃)," and "7 (days)" set as "number of fermentation days (days)."

[0070] 9, for example, "2.0" is set for "burnt smell," "4.0" is set for "sour smell," "3.0" is set for "fruity," "10.0 (degrees)" is set for "sugar content (degrees)," "4.0" is set for "pH," "51 (degrees)" is set for "brewing kettle temperature (degrees)," and "10 (days)" is set for "fermentation days." Explanation of the other data included in FIG. 9 will be omitted.

[0071] That is, in this case, the model generation unit 113 generates a second learning model MD2 that can output (predict) values ​​corresponding to, for example, "burnt smell," "sour smell," and "fruity," etc., in response to input of values ​​corresponding to, for example, "sugar content (degrees)," "pH," "brewing kettle temperature (℃)," and "number of fermentation days (days)."

[0072] Returning to FIG. 7, the model generation unit 113 generates a third learning model MD3 by machine learning of the third teacher data DT13 (e.g., the third teacher data DT13 generated in advance by the data generation unit 112) stored in the information storage area 130 (S14). Note that S14 may be performed in parallel with S12 and S13, for example. Specific examples of the third teacher data DT13 will be described below.

[0073] [Example of the third training data DT13] FIG. 10 is a diagram illustrating a specific example of the third teacher data DT13.

[0074] The third teacher data DT13 shown in Figure 10 has, for example, items such as "sugar content (degrees)" and "pH" for which the type of analytical value data DT3 is set, and "brewing kettle temperature (℃)" and "number of fermentation days (days)" for which the type of process data DT2 is set.

[0075] Specifically, the third training data DT13 in the first row shown in Figure 10 has, for example, the "sugar content (degrees)" set to "11.0 (degrees)", the "pH" set to "4.0", the "brewing kettle temperature (℃)" set to "62 (℃)", and the "number of fermentation days (days)" set to "7 (days)".

[0076] 10, for example, "10.0 (degrees)" is set as "sugar content (degrees)", "4.0" is set as "pH", "51 (degrees)" is set as "brewing kettle temperature (degrees)", and "10 (days)" is set as "fermentation days". Explanation of other data included in FIG. 10 will be omitted.

[0077] In other words, in this case, the model generation unit 113 generates a third learning model MD3 that can output (predict) values ​​corresponding to, for example, "brewing kettle temperature (℃)" and "number of fermentation days (days)" when values ​​corresponding to, for example, "sugar content (degrees)" and "pH" are input.

[0078] In the above example, a case was described in which a third learning model MD3 is generated that outputs values ​​corresponding to "brewing kettle temperature (°C)" and "number of fermentation days (days)" in response to input of values ​​corresponding to "sugar content (degrees)" and "pH", etc., respectively, but this is not limited to this. Specifically, the model generation unit 113 may be configured to generate a third learning model MD3 that can output (predict) values ​​corresponding to "sugar content (degrees)" and "pH", etc. in response to input of values ​​corresponding to "brewing kettle temperature (°C)" and "number of fermentation days", etc., respectively.

[0079] In the above example, the first learning model MD1, the second learning model MD2, and the third learning model MD3 are successively generated in the model generation process, but this is not limited to this. The model generation unit 113 may generate, for example, at least one of the first learning model MD1, the second learning model MD2, and the third learning model MD3 in the model generation process. In other words, the model generation unit 113 may generate, for example, the first learning model MD1, the second learning model MD2, and the third learning model MD3 at different times.

[0080] [Information Output Processing in the First Embodiment] Next, the information output process in the first embodiment will be described. Fig. 11 is a flowchart illustrating the information output process in the first embodiment. Fig. 12 and Fig. 13 are diagrams illustrating the information output process in the first embodiment.

[0081] 11, the type identification unit 114 waits, for example, until it is time to output information (NO in S21). The information output time may be, for example, the time when the brewing engineer inputs information indicating that the information output process will be performed.

[0082] Then, when the information output timing arrives (YES in S21), the type identification unit 114 identifies, for example, a first type whose contribution to the overall evaluation data DT4 in the first learning model MD1 satisfies a first condition from among the types of the sensory evaluation data DT1 included in the first teacher data DT11 stored in the information storage area 130 (S22). Note that hereinafter, the function of the type identification unit 114 that performs S22 is also referred to as a first type identification unit.

[0083] Thereafter, the information output unit 115 outputs, for example, result information indicating the first type identified in S22 to the output device (not shown) of the operation terminal 5 (S23). A specific example of the screen data OP1 output in S23 will be described below.

[0084] [Example of screen data OP1] FIG. 12 is a diagram illustrating a specific example of the screen data OP1.

[0085] Specifically, in S22, when "burnt smell", "sour taste", "fruity", "grassiness", and "grain smell" are identified as the first type among the types of sensory evaluation data DT1 contained in the first teacher data DT11 stored in the information storage area 130, the information output unit 115 generates and outputs screen data OP1 including, for example, a bar graph showing the magnitude of contribution of each of "burnt smell", "sour taste", "fruity", "grassiness", and "grain smell" (for example, a bar graph sorted in descending order of contribution), as shown in FIG. 12.

[0086] 11, the type identification unit 114 identifies a second type (S24) in which the contribution of the first type (the first type identified in S22) in the second learning model MD2 to the sensory evaluation data DT1 satisfies a second condition, for example, from at least one of the types of the process data DT2 and the analysis value data DT3 included in the second teacher data DT12 stored in the information storage area 130. Note that hereinafter, the function of the type identification unit 114 that performs S24 is also referred to as the second type identification unit.

[0087] Thereafter, the information output unit 115 outputs, for example, result information indicating the second type identified in S24 to the output device of the operation terminal 5 (S25). A specific example of the screen data OP2 output in S25 will be described below.

[0088] [Example of screen data OP2] FIG. 13 is a diagram illustrating a specific example of the screen data OP2.

[0089] Specifically, in S22, when "number of fermentation days (days)", "brewing kettle temperature (℃)", and "number of yeast storage days (days)" among the types of process data DT2 contained in the second teacher data DT12 stored in the information storage area 130 are identified as part of the second type, and "sugar content (degrees)" and "pH" among the types of analytical value data DT3 contained in the second teacher data DT12 stored in the information storage area 130 are identified as other parts of the second type, the information output unit 115 generates and outputs screen data OP2 including, for example, a bar graph (for example, a bar graph sorted in descending order of contribution) showing the magnitude of contribution of each of "number of fermentation days (days)", "brewing kettle temperature (℃)", "number of yeast storage days (days)", "sugar content (degrees)", and "pH", as shown in Figure 13.

[0090] In other words, in this case, the information output unit 115 generates and outputs screen data OP2 that mixes, for example, contribution degrees corresponding to the type of process data DT2 and contribution degrees corresponding to the type of analysis value data DT3, which are mutually different data.

[0091] In addition, in S24, the type identification unit 114 may identify as the second type, for example, a type among at least one of the process data DT2 and the analysis value data DT3 contained in the second teacher data DT12 stored in the information storage area 130, whose contribution to the sensory evaluation data DT1 of a type other than the first type in the second learning model MD satisfies a second condition.

[0092] 11, the type identification unit 114 identifies, for example, a type (hereinafter also referred to as the third type) of the process data DT2 included in the third teacher data DT13 stored in the information storage area 130, in which the contribution of the second type (the second type identified in S23) in the third learning model MD3 to the analysis value data DT3 satisfies a condition (hereinafter also referred to as the third condition) (S26). Note that, hereinafter, the function of the type identification unit 114 that performs S26 is also referred to as the third type identification unit.

[0093] Specifically, in S26, the type identification unit 114 identifies, for example, among the types of process data DT2 included in the third teacher data DT13, a type whose contribution to the second type of analysis value data DT3 in the third learning model MD3 is greater than or equal to a predetermined threshold (hereinafter also referred to as the third threshold) as the third type.

[0094] In addition, in S26, the type identification unit 114 identifies, for example, a predetermined number (hereinafter also referred to as the third predetermined number) of types of process data DT2 included in the third teacher data DT13 that have the highest contribution to the second type of analysis value data DT3 in the third learning model MD3 as the third type.

[0095] Thereafter, the information output unit 115 outputs, for example, result information indicating the third type identified in S26 to the output device of the operation terminal 5 (S27).

[0096] Specifically, the information output unit 115 outputs to the output device of the operation terminal 5 screen data (not shown) including a bar graph in which the third types identified in S26 are sorted in descending order of contribution.

[0097] In this way, in the model generation process, the information processing device 1 in this embodiment generates a first learning model MD1 that can output (predict) overall evaluation data DT4 in response to input of sensory evaluation data DT1, for example, by machine learning using multiple first teacher data DT11 each including sensory evaluation data DT1 and overall evaluation data DT4.

[0098] In addition, in the model generation process, the information processing device 1 in this embodiment generates a second learning model MD2 that can output (predict) sensory evaluation data DT1 in response to input of at least one of the process data DT2 and the analysis value data DT3, for example, by machine learning using multiple second teacher data DT12 each including at least one of the process data DT2 and the analysis value data DT3 and the sensory evaluation data DT1.

[0099] Furthermore, in the information output process, the information processing device 1 in this embodiment identifies, for example, a first type among the types of sensory evaluation data DT1 included in the first teacher data DT11, whose contribution to the overall evaluation data DT4 in the first learning model MD1 satisfies a first condition. Then, the information processing device 1 in this embodiment identifies, for example, a second type among at least any of the types of process data DT2 and analysis value data DT3 included in the second teacher data DT12, whose contribution to the sensory evaluation data DT1 of the first type in the second learning model MD2 satisfies a second condition. Thereafter, the information processing device 1 in this embodiment outputs, for example, information indicating each of the identified second types.

[0100] That is, the information processing device 1 in this embodiment, for example, identifies a first type of sensory evaluation data DT1 that has a high contribution to overall evaluation data DT4 in a first learning model MD1 for the target beverage as the type of sensory evaluation data DT1 that is important in bringing the flavor of the target beverage closer to a predetermined flavor (in other words, a target flavor).The information processing device 1 in this embodiment then outputs, for example, a second type of process data DT2 or analytical value data DT3 that has a high contribution to the first type of sensory evaluation data DT1 in a second learning model MD2 for the target beverage as the type of process data DT2 or the type of analytical value data DT3 that is important in adjusting the first type of sensory evaluation data DT1.

[0101] As a result, the information processing device 1 in this embodiment can output, for example, the type of process data DT2 and the type of analytical value data DT3, which are important when bringing the flavor of the target beverage closer to a specified flavor, as result information.

[0102] Therefore, a brewing engineer can efficiently and stably make adjustments to maintain or improve the flavor of the target beverage by, for example, referring to the information output by the information processing device 1. Specifically, a brewing engineer can efficiently review the brewing process to maintain or improve the flavor of the target beverage.

[0103] In particular, the information processing device 1 in this embodiment is capable of outputting information indicating types of process data DT2 or analytical value data DT3 that are important when bringing the flavor of the target beverage closer to a specified flavor, but that brewing engineers were not aware of (in other words, information indicating knowledge that is not generally known).

[0104] Therefore, by referring to information output by the information processing device 1, for example, about aspects that the brewing technician was not aware of, the brewing technician can make adjustments to maintain or improve the flavor of the target beverage more efficiently and stably.

[0105] The brewing engineer can then brew the target beverage based on the information output by the information processing device 1. Specifically, the brewing engineer can adjust the production conditions in the second type of process data, for example, to produce a target beverage with an improved flavor.

[0106] The brewing engineer may perform the model generation process and the information output process at regular intervals (hereinafter also referred to as predetermined periods), for example, once a month.

[0107] That is, for example, a brewing engineer may acquire, for each predetermined period, sensory evaluation data DT1, process data DT2, analytical value data DT3, and overall evaluation data DT4 for a target beverage produced within each period. Then, for each predetermined period, the brewing engineer may generate a first learning model MD1, a second learning model MD2, and a third learning model MD3 corresponding to each period by using the sensory evaluation data DT1, process data DT2, analytical value data DT3, and overall evaluation data DT4 acquired during each period, and further acquire result information corresponding to each period.

[0108] This allows the brewing engineer to repeatedly review the brewing process for the target beverage, for example, depending on the state of the target beverage produced in the most recent period, thereby enabling the brewing engineer to review the brewing process for the target beverage with greater accuracy.

[0109] Furthermore, for example, when reviewing the brewing process for a target beverage, a brewing engineer may appropriately perform a predictive process (inference process) using the first learning model MD1, the second learning model MD2, and the third learning model MD3. Specifically, for example, when checking changes in the overall evaluation data DT4 after reviewing the brewing process, a brewing engineer may appropriately perform a predictive process using the second learning model MD2 and the third learning model MD3. [Explanation of symbols]

[0110] 1: Information processing device 5: Operation terminal 100: Information Processing System 101: CPU 102: Memory 103: Communication device 104: Storage medium 105: Bus 110: Program 111: Information Management Department 112: Data generation unit 113: Model generation unit 114: Type identification unit 115: Information output unit 130: Information storage area DT1: Sensory evaluation data DT2: Process data DT3: Analysis value data DT4: Comprehensive evaluation data DT11: First training data DT12: Second training data DT13: Third training data MD1: First learning model MD2: Second learning model MD3: Third learning model NW: Network OP1: Screen data OP2: Screen data

Claims

1. a first model generation unit that generates a first learning model by machine learning using a plurality of first teacher data including sensory evaluation data of a brewed beverage and comprehensive evaluation data indicating a comprehensive evaluation of the flavor of the brewed beverage with respect to a predetermined flavor; an information processing device having a second model generation unit that generates a second learning model by machine learning using a plurality of second teacher data including either process data related to the brewing process of the brewed beverage or analytical value data of the brewed beverage, and the sensory evaluation data.

2. the second model generation unit generates the second learning model by machine learning using the plurality of second teacher data including the process data, the analysis value data, and the sensory evaluation data. The information processing device according to claim 1 .

3. further generating a third learning model by machine learning using a plurality of third teacher data including the process data and the analysis value data; The information processing device according to claim 1 .

4. a memory unit that stores a first learning model generated using a plurality of first teacher data including sensory evaluation data of a brewed beverage and overall evaluation data indicating an overall evaluation of the flavor of the brewed beverage with respect to a predetermined flavor; and a second learning model generated using a plurality of second teacher data including either process data related to a brewing process of the brewed beverage or analytical value data of the brewed beverage and the sensory evaluation data; a first type identification unit that identifies a first type, among the types of the sensory evaluation data included in the first teacher data, whose contribution to the comprehensive evaluation data in the first learning model satisfies a first condition; a second type identification unit that identifies a second type, from among the types of the process data and the analysis value data included in the second teacher data, whose contribution rate to the sensory evaluation data of the first type in the second learning model satisfies a second condition; an information output unit that outputs information indicating the specified second type.

5. the information output unit outputs information indicating the first type. The information processing device according to claim 4 .

6. the first type identification unit identifies, as the first type, a type of the sensory evaluation data whose contribution to the comprehensive evaluation data in the first learning model is equal to or greater than a first threshold, from among the types of the sensory evaluation data; the second type identification unit identifies, as the second type, one of the types of the process data and the analysis value data included in the second teacher data, the type whose contribution rate to the sensory evaluation data of the first type in the second learning model is equal to or greater than a second threshold value; The information processing device according to claim 4 .

7. the first type identification unit identifies, as the first types, a first predetermined number of types from among the types of the sensory evaluation data, which types have a high degree of contribution to the comprehensive evaluation data in the first learning model; the second type identification unit identifies, as the second types, a second predetermined number of types of the process data and the analysis value data included in the second teacher data, in descending order of contribution to the sensory evaluation data of the first type in the second learning model; The information processing device according to claim 4 .

8. The storage unit stores a third learning model generated by machine learning using a plurality of third teacher data including the process data and the analysis value data, and further a third type identification unit that identifies a third type among the types of the process data included in the third teacher data, the third type having a contribution rate to the analysis value data of the second type in the third learning model that satisfies a third condition; the information output unit outputs information indicating the identified third type. The information processing device according to claim 4 .

9. The information processing device according to claim 4 , wherein the first learning model and the second learning model are a decision tree model or a random forest model.

10. A method for producing a brewed beverage, comprising producing a brewed beverage based on production conditions in the second type of process data identified by the information processing device according to claim 4 .

11. generating a first learning model by machine learning using a plurality of first teacher data including sensory evaluation data of the brewed beverage and comprehensive evaluation data indicating a comprehensive evaluation of the flavor of the brewed beverage with respect to a predetermined flavor; generating a second learning model by machine learning using a plurality of second teacher data including either process data related to a brewing process of the brewed beverage or analytical value data of the brewed beverage, and the sensory evaluation data; A learning model generation method in which processing is performed by a computer.

12. a first learning model generated using a plurality of first teacher data including sensory evaluation data of a brewed beverage and overall evaluation data indicating an overall evaluation of the flavor of the brewed beverage relative to a predetermined flavor; and a second learning model generated using a plurality of second teacher data including either process data related to a brewing process of the brewed beverage or analytical value data of the brewed beverage, and the sensory evaluation data, in a storage unit; Identifying a first type of the sensory evaluation data included in the first teacher data, the contribution of which to the overall evaluation data in the first learning model satisfies a first condition; Identifying a second type of data among the process data and the analysis value data included in the second teacher data, the second type of data having a contribution rate to the sensory evaluation data of the first type in the second learning model that satisfies a second condition; outputting information indicating the identified second type; An information output method in which processing is performed by a computer.

13. An information display method for displaying types of process data relating to the brewing process of a brewed beverage and types of analytical value data of the brewed beverage in a mixed manner in descending order of contribution to sensory evaluation data of the brewed beverage.

Images