Issuing system, program, and issuing method

Abstract

To provide an issuing system, a program and an issuing method.SOLUTION: An issuing system that issues a quality warranty certificate to a user in response to a request for the quality warranty certificate for a target food from the user, the system comprising: a receiving unit that receives a request for the quality warranty certificate for the target food from the user; a measurement result acquiring unit that acquires a measurement result of the target food measured by a sensor; and an issuing unit that inputs the measurement result acquired by the measurement result acquiring unit to any learning model of a plurality of types of learning models that take the measurement result of the food as an input and outputs an evaluation of a plurality of taste-related items associated with the taste of the food as an output, and issues the quality warranty certificate including the evaluation acquired by the evaluation acquiring unit to the user.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an issuing system, a program, and an issuing method. [Background technology]

[0002] Patent Document 1 describes a quality assessment method and system for fresh food. Patent Document 2 describes a commercial transaction system that guarantees the quality of goods. Patent Document 3 describes a comprehensive grading and display system for quality assurance of semiconductor assembly products. [Prior art document] [Patent documents] [Patent Document 1] JP 2024-027765 A [Patent Document 2] International Publication No. 2007 / 129488 [Patent Document 3] Japanese Patent Application Laid-Open No. 2002-286797 Summary of the Invention [Means for solving the problem]

[0003] According to one embodiment of the present invention, an issuing system is provided. The issuing system may include a receiving unit that receives a user's request for a quality assurance certificate for a target food product. The issuing system may include a measurement result acquisition unit that acquires measurement results obtained by a sensor measuring the target food product. The issuing system may include an evaluation acquisition unit that inputs the measurement results of the food product acquired by the measurement result acquisition unit into one of multiple types of learning models that input evaluations of multiple taste-related items related to the taste of the food product, and acquires the evaluation output from the learning model. The issuing system may include an issuing unit that issues the quality assurance certificate to the user, including the evaluation acquired by the evaluation acquisition unit.

[0004] The issuing system may include a storage unit that stores the learning model for each type of food and each place of production or producer. The evaluation acquisition unit may input the measurement results into the learning model corresponding to the type of food and the place of production or producer of the target food, and acquire the evaluation output from the learning model.

[0005] The issuing system may include a storage unit that stores the learning model for each type of food and for each production area or producer. The evaluation acquisition unit may input the measurement results into a learning model specified by the user from among multiple types of learning models stored in the model storage unit, and acquire the evaluation output from the learning model.

[0006] Any of the issuance systems may include a learning data acquisition unit that acquires learning data including measurement results of food by a sensor and sensory test results of the food, the type of food, and the place of production or producer of the food, and a learning model creation unit that performs machine learning using the multiple pieces of learning data acquired by the learning data acquisition unit to create a learning model corresponding to the type acquired by the learning data acquisition unit and at least one of the place of production and the producer, and the evaluation acquisition unit may use any of the multiple types of learning models created by the learning model creation unit.

[0007] In any of the above-mentioned issuing systems, the issuing unit may issue the quality assurance certificate including used learning model identification information that can identify the learning model used by the evaluation acquisition unit from among the multiple types of learning models.

[0008] In any of the issuing systems, the issuing unit may issue the quality assurance certificate including an assurance level that indicates a higher level the higher the degree of relevance between the learning model used by the evaluation acquisition unit from among the multiple types of learning models and the target food.

[0009] In any of the issuing systems, the multiple types of learning models may take the measurement results of the food as input and output a measurement evaluation which is an evaluation of the multiple taste-related items at the time the food is measured by the sensor, and a future evaluation which is an evaluation of the multiple taste-related items in the future at the time the food is measured by the sensor, the evaluation acquisition unit may input the measurement results acquired by the measurement result acquisition unit into any of the multiple types of learning models and acquire the measurement evaluation and the future evaluation output from the learning model, and the issuing unit may issue the quality assurance certificate to the user which includes the measurement evaluation and the future evaluation acquired by the evaluation acquisition unit.

[0010] In any of the above-mentioned issuing systems, the issuing unit may issue the quality assurance certificate including at least one of a recommended cooking method for the food and a recommended cooking time for the food, based on the evaluation of the target food obtained by the evaluation acquisition unit.

[0011] In any of the issuing systems, the issuing unit may issue the quality assurance certificate according to the recipient of the target food.

[0012] In any of the issuing systems described above, the measurement result acquisition unit may acquire measurement results obtained by measuring the target food with a spectroscopic sensor.

[0013] According to one embodiment of the present invention, there is provided an issuing method executed by a computer. The issuing method may include a receiving step of receiving a user's request for a quality assurance certificate for a target food product. The issuing method may include a measurement result acquisition step of acquiring measurement results of the target food product measured by a sensor. The issuing method may include an evaluation acquisition step of inputting the measurement results acquired in the measurement result acquisition step into one of multiple types of learning models that input the food measurement results and output evaluations of multiple taste-related items related to the taste of the food, and acquiring an evaluation output from the learning model. The issuing method may include an issuing step of issuing the quality assurance certificate including the evaluation acquired in the evaluation acquisition step to the user.

[0014] According to one embodiment of the present invention, there is provided a program for causing a computer to execute the issuing method.

[0015] The above summary of the invention does not list all of the necessary features of the present invention, and subcombinations of these features may also constitute inventions. [Brief explanation of the drawings]

[0016] [Figure 1] 1 illustrates a schematic diagram of an example issuing system 100. [Figure 2] 1 shows a schematic diagram of multiple types of learning models 200 stored in the issuing system 100. [Figure 3] 1 shows a schematic diagram of an example of a quality assurance certificate 104 issued by the issuing system 100. [Figure 4] 1 shows an example of a functional configuration of an issuing system 100. [Figure 5] 10 shows an example of a processing flow by the issuing system 100. [Figure 6] 10 shows an example of a processing flow by the issuing system 100. [Figure 7] 10 shows an example of a processing flow by the issuing system 100. [Figure 8] 10 shows an example of a processing flow by the issuing system 100. [Figure 9] 1 shows an example of a hardware configuration of a computer 1200 that functions as part or all of the issuing system 100. DETAILED DESCRIPTION OF THE INVENTION

[0017] The present invention will be described below through embodiments of the invention, but the following embodiments do not limit the scope of the invention according to the claims. Furthermore, not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0018] Foods are subject to standards such as sugar content and rank, and wholesale prices are determined based on these standards at the distribution stage. Standards are often determined by subjective human judgment based on the food's appearance or simple measurement results. Conventional food standards are not evaluated or guaranteed in a manner optimized for each local producer (production region). Furthermore, there is no guarantee that quality information will be maintained as objective, secure information without tampering. Furthermore, sugar content and rank (grade) do not necessarily represent quality standards for flavor itself, and flavor often does not match sugar content or rank. Even if food quality standards could be standardized, there is currently no means to disseminate these quality standards to local producers. The issuing system 100 of this embodiment, for example, assesses and evaluates the quality of food corresponding to its flavor and issues quality certificates that can be widely used by local producers (production regions). By issuing a food quality guarantee, the issuing system 100 guarantees the quality of the food, allowing food producers and others to add value through the quality guarantee, which is expected to increase the unit price of food and expand sales channels. In particular, by having producers who are confident in the quality of their products participate in the platform based on the issuing system 100, it will lead to increased added value for food products as a whole.

[0019] FIG. 1 shows a schematic diagram of an example of an issuing system 100. The issuing system 100 may be realized by one device or by multiple devices. For example, the issuing system 100 may be realized by one server. For example, the issuing system 100 may be configured by multiple servers. The issuing system 100 may also be realized by a distributed environment in the cloud.

[0020] The issuing system 100 issues a quality assurance certificate 104 to a user 302 in response to a request 102 by the user 302 for a quality assurance certificate 104 for a target food product 20. The user 302 may be, for example, a producer of the food product 20 or a seller of the food product 20.

[0021] The issuing system 100 stores multiple types of learning models 200 that receive input of measurement results of a food product 20 measured by a sensor 400 and output evaluations of taste-related items related to the taste of the food product 20. The learning models 200 may output evaluations of multiple taste-related items. The taste-related items for which evaluations are output may vary depending on the learning model 200. Examples of taste-related items include bitterness, sweetness, sourness, saltiness, umami, spiciness, astringency, richness, sharpness, freshness, softness, aftertaste, and juiciness. The taste-related items are not limited to these and may be any items related to taste. The multiple taste-related items for which the learning model 200 outputs evaluations may be registered according to the food product 20 targeted by the learning model 200. The multiple taste-related items for which the learning model 200 outputs evaluations may be independent taste-related items that are not correlated with each other. The multiple taste-related items for which the learning model 200 outputs evaluations may be independent taste-related items that are highly uncorrelated with each other. The correlation between taste-related items may be determined according to criteria used in taste research, such as the lack of correlation between bitterness and sweetness. A very low correlation may mean that the degree of correlation is lower than a predetermined threshold. The issuing system 100 may store templates of multiple taste-related items for each food product 20, and the multiple taste-related items actually used for each food product 20 may be determined based on the templates.

[0022] The issuing system 100, for example, receives a request 102 and measurement results obtained by a sensor 400 measuring the target food 20, inputs the received measurement results into one of multiple types of learning models 200, obtains an evaluation output from the learning model 200, and issues a quality assurance certificate 104 including the obtained evaluation.

[0023] The food 20 covered by the issuing system 100 may include all foods and beverages. The food 20 may include fresh food. For example, the food 20 includes agricultural products (rice, cereals, beans, vegetables, fruits, etc.), livestock products (meat, edible avian eggs, etc.), and marine products (fish, shellfish, aquatic animals, marine mammals, seaweed, etc.). The food 20 may include processed foods. For example, food 20 includes wheat, flour, starch, processed vegetables, processed fruit products, tea, coffee and cocoa preparations, spices, noodles, bread, processed grain products, confectionery, prepared beans, sugars, other processed agricultural products, meat products, dairy products, processed egg products, other processed livestock products, processed seafood (dried seafood, salted dried seafood, salted seafood, processed frozen seafood, etc.), processed seaweed (kelp, kelp processed products, dried seaweed, agar, etc.), other processed seafood foods, seasonings and soups (salt, miso, soy sauce, sauces, vinegar, soup, etc.), edible oils and fats (edible vegetable oils and fats, edible animal oils and fats, etc.), cooked foods (cooked frozen foods, chilled foods, retort pouch foods, prepared dishes, etc.), other processed foods, beverages, etc. (drinking water, soft drinks, ice, etc.). Food 20 may include cooked foods.

[0024] The issuing system 100 receives, for example, a request 102 and a measurement result of a target food 20 measured by a sensor 400 from an edge device 300 used by a user 302. The edge device 300 may be a smartphone, a tablet terminal, a PC (Personal Computer), a smart device, or the like. The issuing system 100 may further include an edge device 300.

[0025] The sensor 400 measures components contained in the food 20. The sensor 400 may be a sensor that uses electromagnetic waves. For example, the sensor 400 is a spectroscopic sensor. As an example, the sensor 400 is a near-infrared (NIR) sensor. A spectroscopic sensor separates light into wavelengths and measures the intensity of each wavelength by receiving it with a detector. A characteristic of spectroscopic sensors is that they absorb specific wavelengths of chemicals. The measurement results of the spectroscopic sensor vary depending on the type and amount of chemicals contained in the food 20. In other words, the measurement results of the food 20 obtained by the spectroscopic sensor include information about the components contained in the food 20. Therefore, for example, by performing machine learning using the measurement results of the food 20 and the sensory test results of the food 20 as training data, a learning model can be generated that uses the measurement results of the food 20 as input and outputs content corresponding to the sensory test results of the food 20. The sensor 400 may be any type that, when measuring the food 20, produces different measurement results depending on the components contained in the food 20. For example, the sensor 400 may be an acoustic sensor. For example, the sensor 400 may be an ultrasonic sensor. The issuing system 100 may further comprise a sensor 400 .

[0026] The issuing system 100 may generate the learning model 200 by performing machine learning using multiple pieces of training data including measurement results of the food 20 measured by the sensor 400 and sensory test results of the food 20. The sensory test of the food 20 may, for example, involve a person eating the food 20 and rating it for multiple taste-related items. In the sensory test of the food 20, multiple taste-related items pre-registered for the target food 20 may be presented to the person eating the food 20, and the person eating the food 20 may evaluate the food 20. In the sensory test of the food 20, a template pre-stored for the target food 20 may be presented to a person administering the test, and multiple taste-related items determined by the person administering the test based on the template may be presented to the person eating the food 20, and the person eating the food 20 may evaluate the food 20. In the sensory test of the food 20, a template pre-stored for the target food 20 may be presented to a person eating the food 20, and the person eating the food 20 may evaluate the food 20.

[0027] The issuing system 100 may generate a learning model 200 for each producer of the food 20 for each of the multiple types of food 20, and may generate a learning model 200 for each production area of ​​the food 20. The multiple types of learning models 200 may be hierarchical.

[0028] Figure 2 shows a schematic diagram of multiple types of learning models 200 stored in the issuing system 100. In the example shown in Figure 2, a general-purpose model 210 is located at the top, and an upper layer model 220, a middle layer model 230, and a lower layer model 240 are located below it in this order.

[0029] The general-purpose model 210 may be a general-purpose learning model 200 that targets all foods 20. The upper layer model 220 may be a learning model 200 for each category of food 20. The categories of food 20 may be more specific than fresh foods, processed foods, and cooked foods. For example, the categories of food 20 may include beef, pork, poultry, apples, mandarin oranges, soybean processed products, vegetable processed products, and fruit processed products. The middle layer model 230 may be a further classification of the upper layer model 220. For example, if the upper layer model 220 is a beef cattle learning model 200, the lower middle layer model 230 may be a Wagyu beef learning model 200 or an Angus beef learning model 200. The middle layer model 230 may have multiple layers. For example, below the Wagyu beef learning model 200, there may be a Japanese Black cattle learning model 200, a Japanese Brown cattle learning model 200, a Japanese Polled cattle learning model 200, and a Japanese Shorthorn cattle learning model 200. The number of layers in the middle layer model 230 may vary depending on the category of food 20. The lower layer model 240 may be obtained by further classifying the middle layer model 230 by producer and place of production. Note that the hierarchical structure shown in FIG. 2 is an example, and a different hierarchical structure may be used. For example, the upper layer model 220 may correspond to fresh foods, processed foods, and cooked foods, and the middle layer model 230 may classify these in one or more layers.

[0030] The issuing system 100, for example, generates an upper layer learning model 200 and then generates lower layer learning models 200 from the upper layer model 200. The issuing system 100, for example, creates a general-purpose model 210, creates multiple upper layer models 220 from the general-purpose model 210, creates multiple middle layer models 230 from the upper layer model 220, and creates multiple lower layer models 240 from the middle layer model 230. The issuing system 100 creates the general-purpose model 210, for example, without distinguishing between foods 20, using multiple pieces of training data including measurement results of foods 20 by a sensor 400 and sensory test results of foods 20. The issuing system 100 creates the upper layer model 220 for beef cattle by updating the general-purpose model 210, for example, using multiple pieces of training data including measurement results of beef cattle by a sensor 400 and sensory test results of beef cattle. The issuing system 100 creates a middle layer model 230 for Wagyu beef by updating the upper layer model 220 for beef cattle using a plurality of pieces of learning data including, for example, measurement results of Wagyu beef by the sensor 400 and sensory test results of the Wagyu beef. Then, for example, when the issuing system 100 acquires from the edge device 300 a plurality of pieces of learning data including measurement results of Wagyu beef by the sensor 400 on a certain producer's Wagyu beef and sensory test results of the Wagyu beef from that producer, the issuing system 100 updates the middle layer model 230 for Wagyu beef using the plurality of pieces of learning data to create a lower layer model 240 for the Wagyu beef producer. Also, for example, when the issuing system 100 acquires from the edge device 300 a plurality of pieces of learning data including measurement results of Wagyu beef by the sensor 400 on Wagyu beef from a certain production area and sensory test results of Wagyu beef from that production area, the issuing system 100 updates the middle layer model 230 for Wagyu beef using the plurality of learning data to create a lower layer model 240 for the Wagyu beef production area. When the issuing system 100 creates a new lower layer model 240, it may use it to update the middle layer model 230, use the updated middle layer model 230 to update the upper layer model 220, and use the updated upper layer model 220 to update the general-purpose model 210.

[0031] The issuing system 100, for example, generates a lower-layer learning model 200 and then generates an upper-layer learning model 200 from the lower-layer model 200. The issuing system 100, for example, creates multiple lower-layer models 240, creates a middle-layer model 230 from the multiple lower-layer models 240, creates an upper-layer model 220 from the multiple middle-layer models 230, and creates a general-purpose model 210 from the multiple upper-layer models 220. For example, when the issuing system 100 acquires multiple pieces of learning data from the edge device 300, including measurement results from the sensor 400 of Wagyu beef from a certain producer and sensory test results of the Wagyu beef from the producer, the issuing system 100 uses the multiple pieces of learning data to create a lower-layer model 240 of the Wagyu beef producer. Furthermore, when the issuing system 100 acquires multiple pieces of learning data from the edge device 300, including measurement results from the sensor 400 of Wagyu beef from a certain production area and sensory test results of the Wagyu beef from the production area, the issuing system 100 uses the multiple pieces of learning data to create a lower-layer model 240 of the Wagyu beef production area. For example, when multiple lower layer models 240 for Wagyu beef are created, the issuing system 100 uses the multiple lower layer models 240 for Wagyu beef to create the middle layer model 230 for Wagyu beef. Similarly, the issuing system 100 may create the lower layer model 240 for Angus beef or the middle layer model 230 for Angus beef. The issuing system 100 may create the upper layer model 220 for beef cattle using multiple types of middle layer models 230, such as the middle layer model 230 for Wagyu beef and the middle layer model 230 for Angus beef. Similarly, the issuing system 100 may create the lower layer model 240 for soy sauce, the lower layer model 240 for tofu, the middle layer model 230 for soy sauce, the middle layer model 230 for tofu, and the upper layer model 220 for soybean processed products. When multiple types of upper layer models 220 are created, the issuing system 100 uses the multiple types of upper layer models 220 to create the general-purpose model 210. When the issuing system 100 creates a new lower layer model 240, it may use it to update the middle layer model 230, use the updated middle layer model 230 to update the upper layer model 220, and use the updated upper layer model 220 to update the general-purpose model 210.

[0032] The issuing system 100 may issue a quality assurance certificate 104 using multiple types of learning models 200, including a general-purpose model 210, an upper-layer model 220, a middle-layer model 230, and a lower-layer model 240. For example, the issuing system 100 receives a request 102 for a quality assurance certificate 104 for a target food 20 and measurement results of the target food 20 by a sensor 400 from an edge device 300 of a user 302, selects one of the multiple types of learning models 200, inputs the measurement results into the selected learning model 200, obtains evaluations of multiple taste-related items output from the learning model 200, and issues a quality assurance certificate 104 including the evaluations and used learning model identification information that can identify the learning model 200 used. This makes it possible to issue a quality assurance certificate 104 in which the evaluation of the food 20 is guaranteed by the learning model 200, thereby contributing to improving the added value of the food 20.

[0033] The issuing system 100 may issue a secure quality assurance certificate 104 that has been subjected to tamper-proof processing. The tamper-proof processing used by the issuing system 100 is not particularly limited, and the issuing system 100 may use any existing tamper-proof processing.

[0034] The issuing system 100 may issue a quality assurance certificate 104 including an assurance level. For example, the issuing system 100 issues a quality assurance certificate 104 including an assurance level that indicates a higher level as the relevance between the used learning model 200 and the target food 20 increases.

[0035] The degree of association between the learning model 200 and the target food 20 may be higher as the degree of match between the type of the learning data used to generate the learning model 200 and the target food 20 increases. For example, the degree of association may be highest when the types are a perfect match, such as between Wagyu beef and Wagyu beef, next highest when the hierarchies are different, such as between Wagyu beef and beef cattle, next highest when the types are different in the same hierarchical level, such as between Wagyu beef and Angus cattle, and lowest when the categories do not match. Note that when the categories do not match, the degree of association between Wagyu beef and any of the livestock industries may be higher than between Wagyu beef and any of the industries other than the livestock industry. As a specific example, when the target food 20 is Wagyu beef, the degree of association is highest when using the learning model 200 generated using Wagyu beef training data, the degree of association is next highest when using the learning model 200 generated using beef cattle training data, the degree of association is next highest when using the learning model 200 generated using Angus cattle training data, and the degree of association is lowest when using the learning model 200 generated using training data of other categories. Basically, the higher the degree of match between the learning data used to generate the learning model 200 and the type of target food 20, the higher the accuracy of the evaluation output by the learning model 200. Therefore, by increasing the assurance level as the degree of match between the type is higher, it is possible to achieve an assurance level that conforms to the accuracy of the evaluation output by the learning model 200.

[0036] The degree of association between the learning model 200 and the target food 20 may be higher when the producers of the learning data used to generate the learning model 200 and the target food 20 match, compared to when they do not. For example, if the target food 20 is Wagyu beef from producer A, the degree of association when using a learning model 200 generated using training data on Wagyu beef from producer A is higher than the degree of association when using a learning model 200 generated using training data on Wagyu beef from producer B. Basically, the accuracy of the evaluation output by the learning model 200 is higher when the producers of the training data used to generate the learning model 200 match, compared to when they do not match. Therefore, by increasing the assurance level when the producers match, compared to when they do not match, it is possible to achieve an assurance level that corresponds to the accuracy of the evaluation output by the learning model 200.

[0037] The degree of association between the learning model 200 and the target food 20 may be higher when the production areas of the learning data used to generate the learning model 200 and the target food 20 match, compared to when they do not match. For example, if the target food 20 is Wagyu beef from production area A, the degree of association when using a learning model 200 generated using training data for Wagyu beef from production area A is higher than the degree of association when using a learning model 200 generated using training data for Wagyu beef from production area B. Basically, the accuracy of the evaluation output by the learning model 200 is higher when the production areas of the learning data used to generate the learning model 200 and the target food 20 match, compared to when they do not match. Therefore, by increasing the assurance level when the production areas match, compared to when they do not match, it is possible to achieve an assurance level that corresponds to the accuracy of the evaluation output by the learning model 200.

[0038] The degree of association between learning model 200 and target food 20 may be higher the closer the production area of ​​the learning data used to generate learning model 200 is to target food 20. For example, if production areas B, C, and D are located in order of proximity to production area A, and target food 20 is Japanese beef from production area A, the degree of association is highest when learning model 200 generated using training data of Japanese beef from production area A is used, the degree of association is next highest when learning model 200 generated using training data of Japanese beef from production area B is used, the degree of association is next highest when learning model 200 generated using training data of Japanese beef from production area C is used, and the degree of association is next highest when learning model 200 generated using training data of Japanese beef from production area D is used. Basically, the closer the production area of ​​the learning data used to generate learning model 200 is to target food 20, the higher the accuracy of the evaluation output by learning model 200 tends to be. Therefore, by increasing the assurance level the closer the production location, the assurance level can be set to correspond to the accuracy of the evaluation output by the learning model 200.

[0039] The issuing system 100 may select the learning model 200 to use in accordance with instructions from the user 302. For example, if the user 302 is a producer of the food product 20, and a lower layer model 240 corresponding to the user 302 for the target food product 20 is stored in the issuing system 100, the user 302 selects the lower layer model 240. For example, if the user 302 is a producer of the food product 20, and a lower layer model 240 corresponding to the production area of ​​the target food product 20 is stored in the issuing system 100, the user 302 selects the lower layer model 240. As a result, the user 302 who created the lower layer model 240 corresponding to the user 302 can be issued a quality assurance certificate 104 with a high level of assurance, thereby increasing the added value of the food product 20.

[0040] For example, if user 302 is a producer of food 20 and a lower layer model 240 corresponding to the user 302 for the target food 20 is not stored in the issuing system 100, user 302 selects a learning model 200 with a higher assurance level from among the multiple types of learning models 200. For example, if user 302 is a producer of Wagyu beef and neither a lower layer model 240 corresponding to the user 302 nor a lower layer model 240 corresponding to the user 302's place of production is stored in the issuing system 100, user 302 may select a lower layer model 240 corresponding to another Wagyu beef producer or a lower layer model 240 corresponding to another place of production of Wagyu beef. Also, for example, user 302 may select the middle layer model 230 for Wagyu beef. Although this lowers the assurance level compared to when user 302 selects the lower layer model 240 corresponding to the user 302, it is still possible to obtain a certain level of assurance and increase the added value of the food 20.

[0041] As described above, according to the issuing system 100 of this embodiment, for example, if the user 302 is a producer, and a lower layer model 240 corresponding to the user is stored in the issuing system 100, the user can obtain a quality assurance certificate 104 with a high level of assurance. This can motivate the user 302 to prepare the measurement results of the sensor 400 of the user's food 20 and the sensory test results of the food 20, and to generate a lower layer model 240 corresponding to the user. This can increase the overall amount of training data, and can improve the accuracy of the middle layer model 230, the upper layer model 220, and the general-purpose model 210.

[0042] When a new type of food 20 is added as a target, the issuing system 100 may use a higher-level learning model 200 to generate a learning model 200 for the food 20. For example, when a new soybean processed product is added as a target, the issuing system 100 uses a higher-level model 220 for the soybean processed product to generate a learning model 200 for the new soybean processed product. For example, when a new category of food 20 is added as a target, the issuing system 100 uses a general-purpose model 210 to generate a learning model 200 for the food 20 of that category. By improving the accuracy of the general-purpose model 210 and the higher-level model 220 in advance, the accuracy of the new learning model 200 generated from the general-purpose model 210 and the higher-level model 220 will also be improved, which can contribute to an overall improvement in the accuracy of quality assurance by the issuing system 100.

[0043] The issuing system 100 may automatically select the learning model 200 to be used, rather than selecting the learning model 200 to be used according to instructions from the user 302. For example, the issuing system 100 automatically selects the learning model 200 to be used so as to provide a higher level of assurance. For example, when the user 302 is the producer of the food 20, if the issuing system 100 stores a lower layer model 240 corresponding to the user 302 of the target food 20 or a lower layer model 240 corresponding to the place of production of the target food 20, the issuing system 100 selects the lower layer model 240. This allows the quality assurance certificate 104 with the highest level of assurance to be issued. If the issuing system 100 does not store a lower layer model 240 corresponding to the user 302 of the target food 20 or a lower layer model 240 corresponding to the place of production of the target food 20, the issuing system 100 selects a lower layer model 240 corresponding to another producer of the target food 20, a lower layer model 240 corresponding to another place of production of the target food 20, or a middle layer model 230 corresponding to the target food 20. This allows issuing a quality guarantee certificate 104 with a higher guarantee level, although not the highest, which can contribute to improving the added value of food 20 of user 302.

[0044] The issuing system 100 may issue a quality assurance certificate 104 that includes a measurement evaluation, which is an evaluation of multiple taste-related items at the time of measurement of the target food 20, and a future evaluation, which is an evaluation of multiple taste-related items in the future from the time of measurement. For example, for a food 20 that is measured by a sensor 400 at the time of shipment, the issuing system 100 performs machine learning using multiple pieces of training data including the measurement results of the food 20 measured by the sensor 400 at the time of shipment of the food 20, the sensory test results of the food 20 at the time of shipment, and the sensory test results of the food 20 in the future from the time of shipment, to generate a lower layer model 240 that takes the measurement results of the food 20 at the time of shipment as input and outputs the evaluation of multiple taste-related items of the food 20 at the time of shipment and the evaluation of multiple taste-related items in the future from the time of shipment of the food 20. The issuing system 100 then inputs the measurement results of the target food 20 at the time of shipment into the lower layer model 240, obtains the measurement evaluation and future evaluation from the lower layer model 240, and includes them in the quality assurance certificate 104. This makes it possible to grasp the evaluation at the time of shipment as well as the evaluation at a certain time after the time of shipment, making it easier to select food 20 that meets the customer's needs, to determine where and how to store food 20, and to consider how to cook food 20.

[0045] The issuing system 100 may issue a quality assurance certificate 104 that includes a recommended cooking method for the target food 20 based on the evaluation of the target food 20. The recommended cooking method for the food 20 may include a recommended way to cook the food 20, or may include a recommended dish menu for the food 20. The issuing system 100 may, for example, pre-store recommended cooking method registration information in which recommended cooking methods for the food 20 are registered for each variation in the evaluation of the taste-related items of the food 20. As a specific example, if the evaluation includes a 5-point rating for each of five taste-related items, namely, item A, item B, item C, item D, and item E, the recommended cooking method registration information may include recommended cooking methods for each variation, such as a recommended cooking method when all items A to E receive a rating of 5 points, a recommended cooking method when only item A receives a rating of 4 points and the others receive a rating of 5 points, etc. The issuing system 100 may input the measurement results of the target food 20 into the learning model 200, obtain an evaluation from the learning model 200, and refer to the recommended cooking method registration information to identify a recommended cooking method corresponding to the obtained evaluation, and include it in the quality assurance certificate 104. This makes it easier to understand the quality of the food 20 when purchasing the food 20, and can also be used as a basis for determining how to cook the food 20, or can make it easier to determine which food 20 to select when the method of cooking the food 20 has been decided.

[0046] The issuing system 100 may issue a quality assurance certificate 104 including a recommended cooking time for the target food 20 based on the evaluation of the target food 20. For example, the issuing system 100 may pre-store recommended cooking time registration information in which a recommended cooking time for the food 20 is registered for each variation in the evaluation of the taste-related items of the food 20. As a specific example, if the evaluation includes a 5-point rating for each of five taste-related items, namely, item A, item B, item C, item D, and item E, the recommended cooking time registration information may include recommended cooking times for each variation, such as a recommended cooking time when all items A to E are rated 5 points, or a recommended cooking time when only item A is rated 4 points and the others are rated 5 points. The issuing system 100 may input the measurement results of the target food 20 into the learning model 200, obtain the evaluation from the learning model 200, and refer to the recommended cooking time registration information to identify a recommended cooking time corresponding to the obtained evaluation and include it in the quality assurance certificate 104. This makes it easier to understand the quality of food 20 when purchasing food 20, for example, and can also be used as a basis for determining when to cook food 20, or, if the time to cook food 20 has been decided, makes it easier to determine which food 20 to select.

[0047] The issuing system 100 may issue a quality assurance certificate 104 according to the recipient to which the target food 20 will be provided. When the issuing system 100 receives a request 102 including the recipient to which the target food 20 will be provided, the issuing system 100 may issue a quality assurance certificate 104 according to the recipient.

[0048] For example, the issuing system 100 issues a quality assurance certificate 104 including a recommended cooking method according to the destination to which the target food 20 is provided. For example, the issuing system 100 pre-stores category-specific recommended cooking method registration information in which recommended cooking methods for the food 20 are registered for each category of destination and for each variation in evaluation of the food 20's taste-related items. The destination category may indicate the genre in which the destination prepares the food 20, such as Japanese cuisine, French cuisine, Italian cuisine, and sushi. The destination category may indicate the attributes of the person receiving the food 20 or a dish made from the food 20, such as gender, age, and nationality. The issuing system 100 may input the measurement results of the target food 20 into the learning model 200, obtain an evaluation from the learning model 200, and refer to the category-specific recommended cooking method registration information corresponding to the destination of the food 20 to identify a recommended cooking method corresponding to the obtained evaluation and include it in the quality assurance certificate 104. This allows, for example, for food 20 used in a French restaurant, to be issued a quality assurance certificate 104 recommending a French cooking method, rather than issuing a quality assurance certificate 104 recommending a Japanese cooking method, thereby improving convenience for user 302.

[0049] For example, the issuing system 100 issues a quality assurance certificate 104 including a recommended cooking time according to the destination to which the target food 20 is provided. For example, the issuing system 100 pre-stores category-specific recommended cooking time registration information in which recommended cooking times for the food 20 are registered for each category of destination and for each variation in the evaluation of the food 20's taste-related items. The issuing system 100 may input the measurement results of the target food 20 into the learning model 200, obtain an evaluation from the learning model 200, and refer to the category-specific recommended cooking time registration information corresponding to the destination of the food 20 to identify a recommended cooking time corresponding to the obtained evaluation and include it in the quality assurance certificate 104. This makes it possible to issue a quality assurance certificate 104 including a recommended cooking time appropriate for the destination, thereby improving convenience for the user 302.

[0050] When the issuing system 100 is realized by one device, the one device may generate and store multiple types of learning models 200. When the issuing system 100 is realized by multiple devices, the multiple types of learning models 200 may be generated and stored in a distributed manner.

[0051] Fig. 3 shows an example of the quality assurance certificate 104 issued by the issuing system 100. Note that the quality assurance certificate 104 shown in Fig. 3 is just an example, and the quality assurance certificate 104 may include only some of the items shown in Fig. 3, or may include items other than those shown in Fig. 3.

[0052] The issuing system 100 may include the target food 20 in the quality assurance certificate 104. FIG. 3 illustrates an example in which the target food 20 is Wagyu beef. The issuing system 100 may include the producer of the target food 20 in the quality assurance certificate 104. The issuing system 100 may include the place of production of the target food 20 in the quality assurance certificate 104. The issuing system 100 may include used learning model identification information that can identify the learning model 200 used in the quality assurance certificate 104. The used learning model identification information may be any information that can identify the learning model 200 used by the issuing system 100. For example, the used learning model identification information may be the name of the learning model 200, as illustrated in FIG. 3. The name of the learning model 200 may be determined based on the learning data used when generating the learning model 200. Figure 3 illustrates an example in which the learning model 200 is generated using training data in which the target food 20 is Wagyu beef and the producer is AAA, and the name of the learning model 200 includes the target food 20, "Wagyu beef," and the producer, "AAA."

[0053] The issuing system 100 may include in the quality assurance certificate 104 an evaluation of the target food 20 at the time of measurement. In the example shown in FIG. 3, the evaluation at the time of measurement includes evaluations of five taste-related items: umami, sweetness, juiciness, softness, and richness. FIG. 3 illustrates an example in which the evaluation is on a five-point scale from 1 star to 5 stars, but this is not limiting. The evaluation may be on a scale other than 5, and the evaluation may be expressed in a form other than stars. The issuing system 100 may include in the quality assurance certificate 104 an assurance level of the evaluation at the time of measurement. In the example shown in FIG. 3, an example in which the assurance level is on 10 levels from 1 to 10, but this is not limiting. The evaluation may be on a scale other than 10, and the evaluation may be expressed in a form other than numerical values. The issuing system 100 may include in the quality assurance certificate 104 a recommended cooking method corresponding to the evaluation at the time of measurement. The issuing system 100 may include a recommended cooking time instead of the recommended cooking method, or may include both the recommended cooking method and the recommended cooking time.

[0054] The issuing system 100 may include a future evaluation of the target food product 20 in the quality assurance certificate 104. In the example shown in FIG. 3, the future evaluation includes evaluations of five taste-related items: umami, sweetness, juiciness, softness, and richness. While FIG. 3 illustrates an example in which the evaluation is on a five-point scale from one star to five stars, this is not limiting. The evaluation may be on a scale other than five, and the evaluation may be expressed in a format other than stars. The quality assurance certificate 104 shown in FIG. 3 includes a future evaluation one month from now, but may include a future evaluation for any period of time, not limited to one month from now. Furthermore, the quality assurance certificate 104 may include multiple future evaluations for different periods of time. The issuing system 100 may include an assurance level for the future evaluation in the quality assurance certificate 104. The issuing system 100 may determine the assurance level for the future evaluation based on the assurance level for the evaluation at the time of measurement and the period for the future evaluation. For example, the issuing system 100 may use the assurance level of the evaluation at the time of measurement as a base point and set the assurance level for the future evaluation to a level that decreases as the period of the future evaluation becomes longer. In the example shown in FIG. 3, the assurance level is 10 levels, from 1 to 10, but this is not limited to this. The evaluation may be on a scale other than 10, and the evaluation may be expressed using a form other than a numerical value. The issuing system 100 may include a recommended cooking method corresponding to the future evaluation in the quality assurance certificate 104. The issuing system 100 may include a recommended cooking time instead of the recommended cooking method, or may include both the recommended cooking method and the recommended cooking time.

[0055] Issuing a quality assurance certificate 104 such as the one shown in Figure 3 makes it possible to easily understand the type, producer, and place of production of the target food 20. Furthermore, since the quality assurance certificate 104 includes used learning model identification information, it becomes possible to verify whether a learning model 200 appropriate for the type, producer, and place of production of the target food 20 has been used. For example, if a learning model 200 for Wagyu beef is used for Wagyu beef, it can be determined that the reliability of the evaluation is higher than when a learning model 200 for Angus beef or a learning model 200 for another category is used.

[0056] By including the evaluation at the time of measurement and the future evaluation in the quality assurance certificate 104, it is possible to understand how the evaluation changes over time in addition to the evaluation at the time of shipping, etc. of the food product 20. For example, if the subject is Wagyu beef, it may have characteristics such as its freshness decreasing in the future compared to the time of measurement, but its flavor and richness increasing as it ages, and it is possible to easily understand such characteristics according to the type of food product 20.

[0057] 4 shows an example of the functional configuration of the issuing system 100. The issuing system 100 includes a storage unit 110, a learning data acquisition unit 112, a learning model generation unit 114, a reception unit 116, a measurement result acquisition unit 118, an evaluation acquisition unit 120, and an issuing unit 122. It is not essential that the issuing system 100 include all of these units.

[0058] The memory unit 110 stores various information. For example, the memory unit 110 stores multiple types of learning models 200. The memory unit 110 may store a learning model 200 for each type of food 20, for each production area or producer. The memory unit 110 may store multiple types of learning models 200 that receive measurement results of the food 20 by the sensor 400 as input and output evaluations of multiple taste-related items of the food 20. The memory unit 110 may store multiple types of learning models 200 that receive measurement results of the food 20 by the sensor 400 as input and output a measurement evaluation that is an evaluation of multiple taste-related items at the time of measurement of the food 20 by the sensor 400 and a future evaluation that is an evaluation of the multiple taste-related items in the future at the time of measurement of the food 20 by the sensor 400. The multiple types of learning models 200 may output future evaluations corresponding to multiple time periods. For example, the multiple types of learning models 200 may output a future assessment one week later, a future assessment two weeks later, a future assessment one month later, a future assessment two months later, etc. The storage unit 110 may store recommended cooking method registration information. The storage unit 110 may store recommended cooking timing information. The storage unit 110 may store recommended cooking method registration information by category. The storage unit 110 may store recommended cooking timing registration information by category.

[0059] The learning data acquisition unit 112 acquires learning data used to generate or update the learning model 200. The learning data acquisition unit 112 may acquire learning data from the edge device 300. The learning data acquisition unit 112 may acquire learning data including measurement results of the food 20 measured by the sensor 400 and sensory test results of the food 20, and the type of the food 20. The learning data acquisition unit 112 may acquire the producer of the food 20. The learning data acquisition unit 112 may acquire the place of production of the food 20.

[0060] The learning model generation unit 114 generates the learning model 200. Generating the learning model 200 may mean generating a new learning model 200, or may include updating an existing learning model 200 to generate a new learning model 200.

[0061] The learning model generation unit 114 performs machine learning using multiple learning data acquired by the learning data acquisition unit 112, and generates a learning model 200 that corresponds to the type of food 20 acquired by the learning model generation unit 114 and at least one of the producer and production area.

[0062] The learning model generation unit 114 may generate a lower layer model 240, generate a middle layer model 230 using the multiple lower layer models 240, generate an upper layer model 220 using the multiple middle layer models 230, and generate a general-purpose model 210 using the multiple upper layer models 220. For example, the learning model generation unit 114 performs machine learning using multiple pieces of training data acquired by the training data acquisition unit 112 to generate a lower layer model 240 corresponding to the type of food 20 acquired by the learning model generation unit 114 and at least one of the producer and production area. The learning model generation unit 114 generates the middle layer model 230 using the multiple lower layer models 240 in response to receiving an instruction from an operator of the issuing system 100 or in response to the number of generated lower layer models 240 exceeding a predetermined number. The learning model generation unit 114 generates the upper layer model 220 using the multiple middle layer models 230 in response to receiving an instruction from an operator of the issuing system 100 or when the number of generated middle layer models 230 exceeds a predetermined number. The learning model generation unit 114 generates the general-purpose model 210 using the multiple upper layer models 220 in response to receiving an instruction from an operator of the issuing system 100 or when the number of generated upper layer models 220 exceeds a predetermined number.

[0063] The learning model generation unit 114 may use the data acquired by the learning data acquisition unit 112 to generate a general-purpose model 210, use the general-purpose model 210 to generate multiple upper layer models 220, use each of the multiple upper layer models 220 to generate multiple middle layer models 230, and use each of the multiple middle layer models 230 to generate multiple lower layer models 240. For example, the learning model generation unit 114 may create the general-purpose model 210 using multiple training data that does not distinguish between foods 20. The learning model generation unit 114 may update the general-purpose model 210 using training data for each category to generate multiple upper layer models 220. The learning model generation unit 114 may update the upper layer model 220 using training data for each type of food 20 to generate multiple middle layer models 230. The learning model generation unit 114 may update the middle layer model 230 using training data for a certain producer's foods 20 to generate a lower layer model 240 for that producer. The learning model generation unit 114 updates the middle layer model 230 using learning data of a food product 20 from a certain production area, thereby generating a lower layer model 240 for that production area.

[0064] When the learning model generation unit 114 creates a new lower layer model 240, it may use the lower layer model 240 to update the middle layer model 230 that is higher than the lower layer model 240, use the updated middle layer model 230 to update the upper layer model 220 that is higher than the updated middle layer model 230, and use the updated upper layer model 220 to update the general-purpose model 210. For example, the learning model generation unit 114 may use the newly generated lower layer model 240 to update the upper middle layer model 230 in response to receiving an instruction from an operator of the issuing system 100 or when the number of generated lower layer models 240 exceeds a predetermined number. The learning model generation unit 114 may use multiple middle layer models 230 to update the upper upper layer model 220 in response to receiving an instruction from an operator of the issuing system 100 or when the number of updates to the middle layer model 230 exceeds a predetermined number. The learning model generation unit 114 updates the general-purpose model 210 using multiple upper layer models 220 in response to receiving instructions from an operator of the issuing system 100 or when the number of updates to the upper layer model 220 exceeds a predetermined number.

[0065] The receiving unit 116 receives a request 102 from a user 302 for a quality assurance certificate 104 for a target food product 20. The receiving unit 116 may receive the request 102 from an edge device 300. The receiving unit 116 may receive the request 102 including a recipient of the target food product 20.

[0066] The measurement result acquisition unit 118 acquires the measurement result obtained by measuring the target food 20 by the sensor 400. If the measurement result is included in the request 102, the measurement result acquisition unit 118 may acquire the measurement result from the request 102. If the measurement result is not included in the request 102, the measurement result acquisition unit 118 may acquire the measurement result separately from the request 102.

[0067] The evaluation acquisition unit 120 inputs the measurement results acquired by the measurement result acquisition unit 118 into one of the multiple types of learning models 200 stored in the storage unit 110 and acquires the evaluation output from the learning model 200. The evaluation acquisition unit 120 may use a learning model 200 specified by the user 302 from the multiple types of learning models 200 stored in the storage unit 110. For example, in response to the reception unit 116 receiving a request 102 from the edge device 300, the evaluation acquisition unit 120 presents the multiple types of learning models 200 stored in the storage unit 110 to the user 302 via the edge device 300 and accepts the designation of the learning model 200. The evaluation acquisition unit 120 may automatically select the learning model 200 to be used. For example, the evaluation acquisition unit 120 selects a learning model 200 corresponding to the type of target food 20 and the producer or production area of ​​the target food 20. When a learning model 200 corresponding to the type, producer, and production area of ​​the target food 20 is stored in the memory unit 110, the evaluation acquisition unit 120 may select the learning model 200. When a learning model 200 corresponding to the type, producer, and production area of ​​the target food 20 is not stored in the memory unit 110, the evaluation acquisition unit 120 may select a learning model 200 corresponding to the type of the target food 20. When a learning model 200 corresponding to the type of the target food 20 is not stored in the memory unit 110, the evaluation acquisition unit 120 may select a learning model 200 corresponding to a higher ranking of the target food 20. As a specific example, when the target food 20 is Wagyu beef, if a learning model 200 corresponding to Wagyu beef and having a matching producer and production area is stored in the memory unit 110, the evaluation acquisition unit 120 may select the learning model 200; otherwise, the evaluation acquisition unit 120 may select a learning model 200 corresponding to Wagyu beef and having a matching producer or production area. If a learning model 200 corresponding to Wagyu beef is not stored in the memory unit 110, the evaluation acquisition unit 120 may select another learning model 200 such as Angus beef cattle from the same hierarchical level, or may select a learning model 200 corresponding to a higher-ranking beef cattle.

[0068] The issuing unit 122 issues the quality assurance certificate 104 including the evaluation acquired by the evaluation acquisition unit 120 to the user 302. For example, when the receiving unit 116 receives the request 102 from the edge device 300, the issuing unit 122 transmits the quality assurance certificate 104 to the edge device 300.

[0069] The issuing unit 122 may include the target food 20 in the quality assurance certificate 104. The issuing unit 122 may include the producer of the target food 20 in the quality assurance certificate 104. The issuing unit 122 may include the place of production of the target food 20 in the quality assurance certificate 104. The issuing unit 122 may include used learning model identification information in the quality assurance certificate 104, which can identify the learning model 200 used by the evaluation acquisition unit 120 from the multiple types of learning models 200 stored in the memory unit 110. The issuing unit 122 may include an assurance level in the quality assurance certificate 104. The issuing unit 122 may include in the quality assurance certificate 104 an assurance level that indicates a higher level the higher the degree of association between the learning model 200 used by the evaluation acquisition unit 120 from the multiple types of learning models 200 and the target food 20.

[0070] The evaluation acquisition unit 120 may input the measurement results acquired by the measurement result acquisition unit 118 into one of multiple types of learning models 200 that output a measurement-time evaluation and a future evaluation, and acquire the measurement-time evaluation and the future evaluation from the learning model 200. The evaluation acquisition unit 120 may acquire a measurement-time evaluation and multiple future evaluations corresponding to multiple periods. The issuing unit 122 may include the measurement-time evaluation and the future evaluation acquired by the evaluation acquisition unit 120 in the quality assurance certificate 104.

[0071] The issuing unit 122 may include at least one of a recommended cooking method and a recommended cooking time for the food 20 in the quality assurance certificate 104 based on the evaluation of the target food 20 acquired by the evaluation acquisition unit 120. The issuing unit 122 may include the recommended cooking method for the target food 20 in the quality assurance certificate 104. The issuing unit 122 may refer to the recommended cooking method registration information stored in the memory unit 110 to identify a recommended cooking method corresponding to the evaluation acquired by the evaluation acquisition unit 120 and include it in the quality assurance certificate 104. The issuing unit 122 may include a recommended cooking time for the target food 20 in the quality assurance certificate 104. The issuing unit 122 may refer to the recommended cooking time registration information stored in the memory unit 110 to identify a recommended cooking time corresponding to the evaluation acquired by the evaluation acquisition unit 120 and include it in the quality assurance certificate 104.

[0072] The issuing unit 122 may issue a quality assurance certificate 104 according to the destination to which the target food 20 is provided. For example, the issuing unit 122 issues a quality assurance certificate 104 including a recommended cooking method according to the destination to which the target food 20 is provided, which is stored in the memory unit 110. The issuing unit 122 may refer to the category-specific recommended cooking method registration information corresponding to the destination to which the food 20 is provided, identify a recommended cooking method corresponding to the evaluation acquired by the evaluation acquisition unit 120, and include the recommended cooking method in the quality assurance certificate 104. For example, the issuing unit 122 issues a quality assurance certificate 104 including a recommended cooking time according to the destination to which the target food 20 is provided. The issuing unit 122 may refer to the category-specific recommended cooking time registration information corresponding to the destination to which the target food 20 is provided, which is stored in the memory unit 110, and identify a recommended cooking time corresponding to the evaluation acquired by the evaluation acquisition unit 120, and include the recommended cooking time in the quality assurance certificate 104.

[0073] 5 shows an example of the flow of processing by the issuing system 100. Here, the flow of processing will be described, in which a request 102 for a quality assurance certificate 104 is received, the quality assurance certificate 104 is issued, and the processing history is stored.

[0074] In step (sometimes abbreviated as S) 102, the receiving unit 116 receives a request 102 for a quality assurance certificate 104 for the target food 20. In S104, the measurement result acquisition unit 118 acquires the measurement results of the sensor 400 measuring the target food 20.

[0075] In S106, the evaluation acquisition unit 120 selects the learning model 200 to be used from the multiple types of learning models 200 stored in the storage unit 110. In S108, the evaluation acquisition unit 120 inputs the measurement results acquired by the measurement result acquisition unit 118 in S104 into the learning model 200 selected in S106, and acquires the evaluation output from the learning model 200.

[0076] In S110, the issuing unit 122 issues a quality assurance certificate 104 including used learning model identification information that can identify the learning model 200 used by the evaluation acquisition unit 120 in S108 and the evaluation acquired by the evaluation acquisition unit 120 in S108.

[0077] In S112, the storage unit 110 stores a processing history including the measurement results acquired by the measurement result acquisition unit 118 in S104 and the evaluation acquired by the evaluation acquisition unit 120 in S108. Then, the processing ends.

[0078] 6 shows an example of the flow of processing by the issuing system 100. Here, we will explain the process of updating the higher-level learning model 200 using the processing history when the processing history satisfies a predetermined condition.

[0079] In S202, the learning model generation unit 114 determines whether the processing history satisfies a predetermined condition. The condition may be, for example, that the number of newly stored processing histories for a certain type of food 20 reaches a predetermined number. If it is determined that the condition is satisfied, the process proceeds to S204.

[0080] In S204, the learning model generation unit 114 uses multiple processing histories to update the upper learning model 200. For example, when the number of processing histories for Wagyu beef reaches a predetermined number, the learning model generation unit 114 uses multiple processing histories to update the learning model 200 corresponding to Wagyu beef (the Wagyu beef middle layer model 230 in FIG. 2). When a measurement result included in the processing history is input to the learning model 200 corresponding to Wagyu beef, the learning model generation unit 114 may update the learning model 200 corresponding to Wagyu beef so that the evaluation output by the learning model 200 corresponding to Wagyu beef approaches the evaluation included in the processing history. When updating the learning model 200 corresponding to Wagyu beef, the learning model generation unit 114 may update the learning model 200 corresponding to beef cattle (the beef cattle upper layer model 220 in FIG. 2), and then update the general-purpose model 210. The output by the learning model 200 corresponding to the lower layer model 240 is considered to be highly accurate because the type, producer, and production area of ​​the food 20 match. By updating the upper level learning model 200 to reflect the output from such a learning model 200 corresponding to the lower layer model 240, it is possible to contribute to improving the accuracy of the upper level learning model 200.

[0081] 7 shows an example of the flow of processing by the issuing system 100. Here, a process of issuing a quality assurance certificate 104 including a measurement evaluation and a future evaluation will be described.

[0082] In S302, the receiving unit 116 receives the request 102 for the quality guarantee certificate 104 for the target food 20. In S304, the measurement result acquiring unit 118 acquires the measurement result of the sensor 400 measuring the target food 20.

[0083] In S306, the evaluation acquisition unit 120 selects the learning model 200 to be used from the multiple types of learning models 200 stored in the storage unit 110. In S308, the evaluation acquisition unit 120 inputs the measurement results acquired by the measurement result acquisition unit 118 in S304 into the learning model 200 selected in S106, and acquires the measurement evaluation and future evaluation output from the learning model 200.

[0084] In S310, the issuing unit 122 issues a quality assurance certificate 104 including used learning model identification information that can identify the learning model 200 used by the evaluation acquisition unit 120 in S308, and the measurement evaluation and future evaluation acquired by the evaluation acquisition unit 120 in S308.

[0085] In S312, the storage unit 110 stores a processing history including the measurement results acquired by the measurement result acquisition unit 118 in S304 and the measurement evaluation and future evaluation acquired by the evaluation acquisition unit 120 in S308. Then, the processing ends.

[0086] 8 shows an example of the flow of processing by the issuing system 100. Here, the process of issuing a quality guarantee certificate 104 according to the recipient of the food product 20 in question will be described.

[0087] In S402, the receiving unit 116 receives a request 102 for a quality assurance certificate 104 for the target food 20, including the recipient of the target food 20. In S404, the measurement result acquisition unit 118 acquires the measurement results of the sensor 400 measuring the target food 20.

[0088] In S406, the evaluation acquisition unit 120 selects the learning model 200 to use from the multiple types of learning models 200 stored in the storage unit 110. In S408, the evaluation acquisition unit 120 inputs the measurement results acquired by the measurement result acquisition unit 118 in S404 into the learning model 200 selected in S406, and acquires the evaluation output from the learning model 200.

[0089] In S410, the issuing unit 122 issues a quality assurance certificate 104 according to the recipient, which includes used learning model identification information that can identify the learning model 200 used by the evaluation acquisition unit 120 in S408 and the evaluation acquired by the evaluation acquisition unit 120 in S408.

[0090] In S412, the storage unit 110 stores a processing history including the measurement results acquired by the measurement result acquisition unit 118 in S404 and the evaluation acquired by the evaluation acquisition unit 120 in S408. Then, the processing ends.

[0091] 9 shows a schematic diagram of an example of the hardware configuration of a computer 1200 that functions as part or all of the publishing system 100. A program installed on the computer 1200 can cause the computer 1200 to function as one or more "parts" of an apparatus according to the present embodiment, or to perform operations associated with the apparatus according to the present embodiment or one or more "parts," and / or to perform a process according to the present embodiment or steps of the process. Such a program can be executed by the CPU 1212 to cause the computer 1200 to perform specific operations associated with some or all of the blocks in the flowcharts and block diagrams described herein.

[0092] The computer 1200 according to this embodiment includes a CPU 1212, a RAM 1214, and a graphics controller 1216, which are interconnected by a host controller 1210. The computer 1200 also includes input / output units such as a communications interface 1222, a storage device 1224, a DVD drive, and an IC card drive, which are connected to the host controller 1210 via an input / output controller 1220. The DVD drive may be a DVD-ROM drive, a DVD-RAM drive, or the like. The storage device 1224 may be a hard disk drive, a solid-state drive, or the like. The computer 1200 also includes a ROM 1230 and legacy input / output units such as a keyboard, which are connected to the input / output controller 1220 via an input / output chip 1240.

[0093] The CPU 1212 operates according to programs stored in the ROM 1230 and the RAM 1214, thereby controlling each unit. The graphics controller 1216 acquires image data generated by the CPU 1212 into a frame buffer or the like provided in the RAM 1214 or into the graphics controller itself, and causes the image data to be displayed on the display device 1218.

[0094] The communication interface 1222 communicates with other electronic devices via a network. The storage device 1224 stores programs and data used by the CPU 1212 in the computer 1200. The DVD drive reads programs or data from a DVD-ROM or the like and provides them to the storage device 1224. The IC card drive reads programs and data from an IC card and / or writes programs and data to an IC card.

[0095] The ROM 1230 stores therein a boot program or the like that is executed by the computer 1200 upon activation, and / or programs that depend on the hardware of the computer 1200. The input / output chip 1240 may also connect various input / output units to the input / output controller 1220 via a USB port, a parallel port, a serial port, a keyboard port, a mouse port, etc.

[0096] The programs are provided by a computer-readable storage medium such as a DVD-ROM or an IC card. The programs are read from the computer-readable storage medium, installed in the storage device 1224, RAM 1214, or ROM 1230, which are also examples of computer-readable storage media, and executed by the CPU 1212. Information processing described in these programs is read by the computer 1200, and causes cooperation between the programs and the various types of hardware resources described above. An apparatus or method may be configured by implementing operations or processing of information in accordance with the use of the computer 1200.

[0097] For example, when communication is performed between the computer 1200 and an external device, the CPU 1212 may execute a communication program loaded into the RAM 1214 and instruct the communication interface 1222 to perform communication processing based on the processing described in the communication program. Under the control of the CPU 1212, the communication interface 1222 reads transmission data stored in a transmission buffer area provided in the RAM 1214, the storage device 1224, a DVD-ROM, or a recording medium such as an IC card, and transmits the read transmission data to the network, or writes reception data received from the network to a reception buffer area or the like provided on the recording medium.

[0098] Furthermore, the CPU 1212 may cause all or a necessary portion of a file or database stored in an external recording medium such as the storage device 1224, a DVD drive (DVD-ROM), an IC card, etc. to be read into the RAM 1214, and may perform various types of processing on the data on the RAM 1214. The CPU 1212 may then write back the processed data to the external recording medium.

[0099] Various types of information, such as various types of programs, data, tables, and databases, may be stored on the recording medium and may undergo information processing. The CPU 1212 may perform various types of processing on data read from the RAM 1214, including various types of operations, information processing, conditional judgment, conditional branching, unconditional branching, information search / replacement, etc., as described throughout this disclosure and specified by the instruction sequences of the programs, and write the results back to the RAM 1214. The CPU 1212 may also search for information in a file, database, etc. on the recording medium. For example, if multiple entries, each having an attribute value of a first attribute associated with an attribute value of a second attribute, are stored on the recording medium, the CPU 1212 may search for an entry whose attribute value of the first attribute matches a specified condition from among the multiple entries, read the attribute value of the second attribute stored in the entry, and thereby obtain the attribute value of the second attribute associated with the first attribute that satisfies a predetermined condition.

[0100] The above-described programs or software modules may be stored in a computer-readable storage medium on or near the computer 1200. A recording medium such as a hard disk or RAM provided in a server system connected to a dedicated communication network or the Internet can also be used as a computer-readable storage medium, thereby providing the programs to the computer 1200 via the network.

[0101] The blocks in the flowcharts and block diagrams in the present embodiments may represent stages of a process in which an operation is performed or "parts" of an apparatus responsible for performing the operation. Particular stages and "parts" may be implemented by dedicated circuitry, programmable circuitry provided with computer-readable instructions stored on a computer-readable storage medium, and / or a processor provided with computer-readable instructions stored on a computer-readable storage medium. The dedicated circuitry may include digital and / or analog hardware circuits, including integrated circuits (ICs) and / or discrete circuits. The programmable circuitry may include reconfigurable hardware circuits, such as field programmable gate arrays (FPGAs) and programmable logic arrays (PLAs), including AND, OR, XOR, NAND, NOR, and other logical operations, flip-flops, registers, and memory elements.

[0102] A computer-readable storage medium may include any tangible device capable of storing instructions that are executed by an appropriate device, such that a computer-readable storage medium having instructions stored thereon comprises an article of manufacture, including instructions that can be executed to create means for performing the operations specified in the flowcharts or block diagrams. Examples of computer-readable storage media may include electronic storage media, magnetic storage media, optical storage media, electromagnetic storage media, semiconductor storage media, etc. More specific examples of computer-readable storage media may include floppy disks, diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), electrically erasable programmable read-only memory (EEPROM), static random access memory (SRAM), compact disc read-only memory (CD-ROM), digital versatile disc (DVD), Blu-ray disc, memory stick, integrated circuit card, etc.

[0103] The computer readable instructions may include either assembler instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state-setting data, or source or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk®, JAVA®, C++, etc., and conventional procedural programming languages ​​such as the “C” programming language or similar programming languages.

[0104] Computer-readable instructions may be provided to a general-purpose computer, a special-purpose computer, or another programmable data processing device, or a programmable circuit, either locally or via a local area network (LAN) or a wide area network (WAN) such as the Internet, so that the processor of the programmable data processing device, such as a computer, or the programmable circuit executes the computer-readable instructions to generate means for performing the operations specified in the flowcharts or block diagrams. Here, the computer may be a personal computer (PC), a tablet computer, a smartphone, a workstation, a server computer, a general-purpose computer, a special-purpose computer, or the like, or may be a computer system in which multiple computers are connected. Such a computer system in which multiple computers are connected is also called a distributed computing system, and is a broad definition of computers. In a distributed computing system, multiple computers collectively execute a program by each executing a portion of the program and passing data between the computers as needed during program execution.

[0105] Examples of processors include computer processors, central processing units, processing units, microprocessors, digital signal processors, controllers, microcontrollers, etc. A computer may have one or more processors. In a multiprocessor system with multiple processors, each processor executes a portion of a program and passes data between processors as needed during program execution, allowing the multiple processors to collectively execute the program. For example, in multitasking, each of the multiple processors may execute a portion of each task in small chunks by switching tasks at time slice intervals. In this case, which portion of a program each processor executes changes dynamically. Which portion of a program each of the multiple processors executes may also be statically determined by multiprocessor-aware programming.

[0106] Although the present invention has been described above using the embodiments, the technical scope of the present invention is not limited to the scope described in the above embodiments. It will be apparent to those skilled in the art that various modifications and improvements can be made to the above embodiments. It is clear from the claims that such modifications and improvements can also be included within the technical scope of the present invention.

[0107] It should be noted that the execution order of each process, such as operations, procedures, steps, and stages, in the devices, systems, programs, and methods shown in the claims, specifications, and drawings is not specifically stated as "before," "prior to," etc., and that the processes can be performed in any order unless the output of a previous process is used in a later process. Even if the operational flow in the claims, specifications, and drawings is described using "first," "next," etc. for convenience, this does not mean that the processes must be performed in this order. [Explanation of symbols]

[0108] 100 Issuing system, 20 Food, 102 Request, 104 Quality assurance certificate, 110 Memory unit, 112 Learning data acquisition unit, 114 Learning model generation unit, 116 Reception unit, 118 Measurement result acquisition unit, 120 Evaluation acquisition unit, 122 Issuance unit, 200 Learning model, 210 General-purpose model, 220 Upper layer model, 230 Middle layer model, 240 Lower layer model, 300 Edge device, 302 User, 400 Sensor, 1200 Computer, 1210 Host controller, 1212 CPU, 1214 RAM, 1216 Graphics controller, 1218 Display device, 1220 Input / output controller, 1222 Communication interface, 1224 Storage device, 1230 ROM, 1240 Input / output chip

Claims

1. a receiving unit that receives a request from a user for a quality assurance certificate for the target food product; a measurement result acquisition unit that acquires measurement results of the target food measured by a sensor; an evaluation acquisition unit that inputs the measurement results of a food product into one of a plurality of learning models that output evaluations of a plurality of taste-related items related to the taste of the food product, and inputs the measurement results acquired by the measurement result acquisition unit into one of a plurality of learning models, and acquires the evaluation output from the learning model; an issuing unit that issues the quality assurance certificate including the evaluation acquired by the evaluation acquisition unit to the user; An issuing system comprising:

2. a storage unit that stores the learning model for each type of food, each place of production, or each producer; Equipped with The issuing system of claim 1, wherein the evaluation acquisition unit inputs the measurement results into the learning model corresponding to the type of the target food and the place of production or producer of the target food, and acquires the evaluation output from the learning model.

3. a storage unit that stores the learning model for each type of food, each place of production, or each producer; Equipped with The issuing system of claim 1, wherein the evaluation acquisition unit inputs the measurement results into a learning model specified by the user from among multiple types of learning models stored in the model memory unit, and acquires the evaluation output from the learning model.

4. a learning data acquisition unit that acquires learning data including the measurement results of the food measured by the sensor and the sensory test results of the food, the type of the food, and the place of production or producer of the food; a learning model creation unit that creates a learning model corresponding to the type acquired by the learning data acquisition unit and at least one of the place of production and the producer by executing machine learning using the plurality of learning data acquired by the learning data acquisition unit; Equipped with The issuing system according to claim 1 , wherein the evaluation acquisition unit uses one of the plurality of types of learning models created by the learning model creation unit.

5. The issuing system according to claim 1 , wherein the issuing unit issues the quality assurance certificate including used learning model identification information that can identify the learning model used by the evaluation acquisition unit from among the plurality of types of learning models.

6. An issuance system described in any one of claims 1 to 5, wherein the issuing unit issues the quality assurance certificate including an assurance level indicating a higher level the higher the degree of relevance between the learning model used by the evaluation acquisition unit among the multiple types of learning models and the target food.

7. The plurality of types of learning models receive the measurement results of the food as input, and output a measurement evaluation that is an evaluation of the plurality of taste-related items at the time of measurement of the food by the sensor, and a future evaluation that is an evaluation of the plurality of taste-related items in the future at the time of measurement of the food by the sensor; the evaluation acquisition unit inputs the measurement results acquired by the measurement result acquisition unit into one of the plurality of types of learning models, and acquires the measurement evaluation and the future evaluation output from the learning model; The issuing system according to claim 1 , wherein the issuing unit issues the quality assurance certificate to the user, the quality assurance certificate including the measurement evaluation and the future evaluation acquired by the evaluation acquisition unit.

8. The issuing system according to any one of claims 1 to 5, wherein the issuing unit issues the quality assurance certificate including at least one of a recommended cooking method for the food and a recommended cooking time for the food based on the evaluation of the target food acquired by the evaluation acquisition unit.

9. The issuing system according to claim 1 , wherein the issuing unit issues the quality assurance certificate according to a recipient of the target food product.

10. The issuing system according to claim 1 , wherein the measurement result acquisition unit acquires the measurement result obtained by measuring the target food with a spectroscopic sensor.

11. 1. A computer-implemented publishing method comprising: a receiving step of receiving a request from a user for a quality assurance certificate for the target food product; a measurement result acquisition step of acquiring a measurement result of the target food measured by a sensor; an evaluation acquisition step of inputting the measurement results acquired in the measurement result acquisition step into one of a plurality of learning models that inputs the measurement results of a food product and outputs evaluations of a plurality of taste-related items related to the taste of the food product, and acquiring the evaluation output from the learning model; an issuing step of issuing the quality assurance certificate including the evaluation acquired in the evaluation acquisition step to the user; An issuing method comprising:

12. A program for causing a computer to execute the issuing method according to claim 11.

Images