Cooking time prediction apparatus

Abstract

To provide a cooking time prediction apparatus capable of quickly and accurately predicting the time required to perform cooking specified in cooking instructions.SOLUTION: A cooking time prediction apparatus 10 comprises: a standardization unit 13 that in a cooking recipe including cooking instructions and information on amounts of ingredients, standardizes names of ingredients included in the cooking instructions to standard ingredient names by referring to an ingredient name table in which a plurality of standard ingredient names are listed, standardizes names of processes included in the cooking instructions to standard process names by referring to a process name table in which a plurality of standard process names are listed, and standardizes the amounts of ingredients included in the cooking recipe to standard ingredient amounts by referring to an ingredient amount table in which a plurality of standard ingredient amounts are listed; and a time required prediction unit 14 that uses the results of machine learning of the accumulated training data to determine a predicted time that is predicted to be required to perform the cooking specified in the cooking instructions.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a cooking time prediction device for predicting cooking time. [Background technology]

[0002] Cooking recipes including cooking instructions that specify the procedures for various processes such as preparing ingredients and heating are published in various media. Some cooking recipes provide an estimated time until the cooking is completed, but it is preferable to know the time required to execute each of the multiple cooking instructions included in the cooking recipe. For example, when there is a cooking instruction such as "boil potatoes," it is preferable to know the time required to execute the cooking instruction. If the time required to execute each cooking instruction is known, for example, when multiple cooking recipes are performed in parallel, a plan can be made in advance to execute a cooking instruction such as "cut carrots" included in one cooking recipe while executing a cooking instruction such as "boil potatoes" included in another cooking recipe.

[0003] Patent Document 1 describes a cooking management device that has a function of specifying a predicted cooking completion time for food provided in a store. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2021-111251 A Summary of the Invention [Problem to be solved by the invention]

[0005] The device described in Patent Document 1 is for determining the predicted cooking completion time for food provided in a store, i.e., for food whose cooking time is approximately known, but is not for predicting the time required to cook unknown foods. In other words, the device described in Patent Document 1 cannot be used to predict the time required to execute, for example, one of the many unknown cooking recipes that are publicly available.

[0006] Therefore, currently, in order to predict the time required to execute the numerous cooking recipes published in various media, a human must look at the cooking instructions for each cooking recipe and make such a prediction based on past experience. However, since there are an infinite number of cooking recipes published, it is extremely cumbersome and unrealistic for a human to make such a time prediction.

[0007] Furthermore, most cooking recipes are written in easy-to-understand language and may contain too much detailed information to predict the time required to perform the cooking steps specified in the cooking instructions. For example, if the cooking instructions are "Stir-fry pork shoulder (thinly sliced)," it is acceptable to simplify the instructions to "cook the thinly sliced ​​meat" in order to predict the time required to perform the cooking step.

[0008] The present invention has been made in consideration of the above-mentioned problems, and its purpose is to provide a cooking time prediction device that can quickly and accurately predict the time required to perform a cooking process specified in a cooking instruction. [Means for solving the problem]

[0009] A characteristic configuration of the cooking time prediction device according to the present invention for achieving the above object is a standardization unit that, for a cooking recipe including information on names of ingredients used in cooking, names of processes related to the cooking of the ingredients, and amounts of the ingredients used in the cooking, refers to an ingredient name table in which a predetermined number of standard ingredient names are listed, and standardizes the names of the ingredients to one of the standard ingredient names, refers to a process name table in which a predetermined number of standard process names are listed, and standardizes the names of the processes to one of the standard process names, refers to an ingredient amount table in which a predetermined number of standard ingredient amounts are listed, and standardizes the amounts of the ingredients to one of the standard ingredient amounts; and a required time prediction unit that uses the results of machine learning execution of accumulated training data to determine a predicted required time that is predicted to be required when the cooking specified in the cooking instruction included in the cooking recipe is performed using ingredients of the standard ingredient names in the standard ingredient amounts standardized by the standardization unit and by processing with the standard process names.

[0010] According to the above characteristic configuration, no matter what wording is used for the names of ingredients, the names of processes, and the amounts of ingredients in the cooking recipe, the standardization unit standardizes them to a predetermined standard ingredient name, a predetermined standard process name, and a predetermined standard ingredient amount. Then, the time required prediction unit can use the results of machine learning on the accumulated training data to determine the predicted time required to perform the cooking specified in the cooking instruction by using ingredients with the standard ingredient names and the standard amounts of ingredients standardized by the standardization unit and processing the standard process names. In particular, in this characteristic configuration, the predicted time required is not determined by a human being, but the time required prediction unit uses the results of machine learning on the accumulated training data to determine the predicted time required to perform the cooking specified in the cooking instruction. As a result, the predicted time required can be obtained in a short time and with consistency. Therefore, it is possible to provide a cooking time prediction device that can quickly and accurately predict the time required to perform a cooking process specified in a cooking instruction.

[0011] Another characteristic configuration of the cooking time prediction device according to the present invention is that the cooking instruction includes a name of a tool used to perform the process, The standardization unit refers to a device name table in which a plurality of standard device names are listed, and standardizes the name of the device included in the cooking instruction to one of the standard device names; The required time prediction unit uses the results of machine learning execution of accumulated training data to determine the predicted required time that is predicted to be required to perform the cooking specified in the cooking instruction using ingredients with the standard ingredient names in the standard ingredient amounts standardized by the standardization unit, and processing with the standard processing name using equipment with the standard equipment name.

[0012] According to the above characteristic configuration, no matter how the name of the utensil used for processing is included in the cooking recipe, it is standardized to a predetermined standard utensil name by the standardization unit. Then, the required time prediction unit can use the results of machine learning on the accumulated training data to determine the estimated required time required to perform the cooking specified in the cooking instruction by processing with the standard processing name using ingredients with the standard ingredient names and the standard amounts of ingredients standardized by the standardization unit, and utensils with the standard utensil names.

[0013] Another characteristic configuration of the cooking time prediction device according to the present invention is that the cooking instruction includes a period-related phrase related to a processing period for performing the processing, The standardization unit standardizes the period-related phrase included in the cooking instruction to a predetermined standard processing period, The required time prediction unit determines the predicted required time taking into account the standard processing period.

[0014] According to the above characteristic configuration, no matter how a time-related term related to the processing period to be processed is contained in a cooking recipe, it is standardized to a predetermined standard processing period by the standardization unit. Here, when a time-related term related to the processing period to be processed is contained in a cooking instruction, the reliability of the estimated time required to perform the cooking specified in the cooking instruction is increased by taking the time-related term into consideration. As a result, the required time prediction unit can determine a highly reliable estimated required time by taking the standard processing period into consideration.

[0015] Another characteristic feature of the cooking time prediction device according to the present invention is that it includes a cooking instruction identification unit that identifies each of the plurality of cooking instructions included in the cooking recipe.

[0016] According to the above characteristic configuration, when one cooking recipe includes multiple cooking instructions, the cooking instruction identification unit can identify each of the multiple cooking instructions included in the cooking recipe. [Brief description of the drawings]

[0017] [Figure 1] FIG. 2 is a diagram showing a configuration of a cooking time prediction device. [Diagram 2] 1 is a diagram illustrating the steps of a cooking time prediction method performed by a cooking time prediction device. FIG. [Diagram 3] FIG. 13 is a diagram showing an example of a material name table. [Figure 4] FIG. 13 is a diagram illustrating an example of a material amount table. [Diagram 5] FIG. 13 is a diagram illustrating an example of a process name table. [Figure 6] FIG. 13 is a diagram showing an example of an instrument name table. [Figure 7] FIG. 13 is a diagram illustrating an example of a period table. [Figure 8] 1 is a graph showing the relationship between predicted cooking time and correct cooking time for a recipe whose cooking time is unknown. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] A cooking time prediction device 10 according to an embodiment of the present invention will be described below with reference to the drawings. The cooking time prediction device 10 of the present embodiment is a device that determines a predicted time required to perform cooking defined in a cooking instruction included in a cooking recipe.

[0019] FIG. 1 is a diagram showing the configuration of a cooking time prediction device 10. The cooking time prediction device 10 includes a standardization unit 13 and a required time prediction unit 14. The cooking time prediction device 10 of this embodiment further includes a cooking recipe acquisition unit 11 and a cooking instruction specification unit 12. The cooking time prediction device 10 is realized using one or more computer devices having an information calculation function, an information input / output function, an information storage function, and the like. In this case, a program (cooking time prediction program) that causes a computer device to realize the functions of the cooking recipe acquisition unit 11, the cooking instruction specification unit 12, the standardization unit 13, and the required time prediction unit 14 may be installed in the computer device.

[0020] In addition, in FIG. 1, the storage device 20 is illustrated separately from the cooking time prediction device 10, but the cooking time prediction device 10 may include the storage device 20.

[0021] In an example described later, the required time prediction unit 14 uses a linear regression model or the like to determine a predicted required time that is predicted to be required to perform a dish specified in a cooking instruction. Furthermore, the required time prediction unit 14 performs machine learning so that, when standardized data created by the standardization unit 13 is input, information related to the predicted required time is output. For example, the coefficients of the linear regression model that constitutes the algorithm of the required time prediction unit 14 (coefficients a1, a2, ..., etc. in Equation 1 described later) are adjusted by machine learning. The results of executing machine learning on the training data are stored and accumulated in the training data storage unit 23 of the storage device 20.

[0022] A cooking recipe storage unit 21 of the storage device 20 stores a plurality of cooking recipes. A cooking recipe includes at least one cooking instruction that explains the procedure for making a target dish. For example, a cooking recipe includes at least one cooking instruction that explains how to prepare each ingredient and how to process each ingredient, arranged in chronological order. Each cooking instruction is composed of a single sentence or multiple sentences. For example, for a dish called potato salad, the cooking recipe includes multiple cooking instructions such as "cut the potatoes," "heat the cut potatoes," and "mash the heated potatoes."

[0023] In addition, a recipe may have an ingredient list that contains information about the amounts of ingredients, for example, "3 potatoes" and "a little salt" in the ingredient list for a potato salad recipe.

[0024] The cooking recipe acquisition unit 11 acquires one of the cooking recipes stored in the cooking recipe storage unit 21. This cooking recipe includes a cooking instruction including the names of ingredients used in the cooking and the names of processes related to cooking the ingredients, as well as information on the amounts of the ingredients. Furthermore, the cooking recipe may also include the names of utensils used in the process and period-related phrases related to the process period.

[0025] The ingredient name includes the name of the ingredient used in the recipe, such as "pork (thinly sliced)" or "salmon (fillet)". Since recipes are created by many different people, they are expressed in a variety of formats. For example, they may be expressed as "pork (thinly sliced)" or "pork shoulder (thinly sliced)" and vary from recipe to recipe.

[0026] The process name includes the name of the process performed in cooking, such as "cut" or "boil." Since cooking recipes are created by many people, they are expressed in various formats. For example, even if the content is the same, it may be expressed as "frying" or simply "heating," and the format may vary from recipe to recipe.

[0027] The names of the utensils include the names of utensils used in cooking, such as "pot" and "frying pan."

[0028] Information on the amount of an ingredient may be, for example, an ingredient list that lists the name of the ingredient and its amount (e.g., weight, number, etc.). The amount of an ingredient also varies from recipe to recipe, and may be expressed in grams (g) or liters (L), in common measuring units such as "1 tablespoon" or "2 cups," or ambiguous expressions such as "a pinch" or "a little," or, if the ingredient is a processed product such as a can, in "1 can."

[0029] The cooking explanation identification unit 12 identifies each of the multiple cooking explanations included in the cooking recipe. For example, the cooking explanation identification unit 12 divides the cooking recipe acquired by the cooking recipe acquisition unit 11 into multiple cooking explanations using punctuation marks, with a combination of a subject and a predicate as a basic unit. Alternatively, the cooking explanation identification unit 12 may identify the cooking recipe acquired by the cooking recipe acquisition unit 11 as a single cooking explanation by using a combination of a predicate (verb) and an object. For example, the cooking explanation identified by the cooking explanation identification unit 12 is stored in the cooking recipe storage unit 21.

[0030] For a cooking recipe including information on the names of ingredients used in cooking, the names of processes related to cooking the ingredients, and the amounts of ingredients used in cooking, the standardization unit 13 refers to an ingredient name table in which a predetermined number of standard ingredient names are listed, and standardizes the names of ingredients included in a cooking instruction including at least the names of ingredients and the names of processes to one of the standard ingredient names; refers to a process name table in which a predetermined number of standard process names are listed, and standardizes the amounts of ingredients included in the ingredient list of the cooking recipe or the cooking instruction to one of the standard ingredient amounts by referring to an ingredient amount table in which a predetermined number of standard ingredient amounts are listed.

[0031] In addition, if the cooking instructions include the name of an utensil used to perform processing related to cooking ingredients, the standardization unit 13 refers to an utensil name table in which multiple standard utensil names are listed, and standardizes the name of the utensil included in the cooking instructions to one of the standard utensil names.

[0032] Furthermore, when the cooking instruction includes a period-related phrase related to a processing period for performing processing related to cooking ingredients, the standardization unit 13 standardizes the period-related phrase included in the cooking instruction to a predetermined standard processing period.

[0033] In the example shown in FIG. 3, which will be described later, standardization refers to a material name table that lists 16 standard material names corresponding to X1 to X16, and converts the name of a material to one of these standard material names.

[0034] In the example shown in FIG. 4, which will be described later, standardization refers to converting the amount of material into one of the units of the standard material amounts (for example, converting it into a numerical value in grams or in units of the number of pieces) by referring to a material amount table that lists eight standard material amounts corresponding to X17 to X24.

[0035] In addition, in the example shown in FIG. 5, which will be described later, standardization refers to a process name table that lists 14 standard process names corresponding to X25 to X38, and converts the name of a process to one of these standard process names.

[0036] In addition, in the example shown in Figure 6 described later, standardization refers to referring to an instrument name table that lists seven standard instrument names corresponding to X39 to X45, and converting the instrument name to one of these standard instrument names.

[0037] In addition, in the example shown in FIG. 7 described later, converting a period-related term into a standard processing period unit corresponding to X46 is called standardization.

[0038] The standardization unit 13 performs extraction processing and standardization processing as described below.

[0039] In the extraction process, the standardization unit 13 extracts the names of ingredients included in the cooking instructions, the names of processes included in the cooking instructions, and the amounts of ingredients included in the cooking recipe. In addition, if the cooking instructions include the names of utensils used to perform the processes, the standardization unit 13 extracts the names of the utensils included in the cooking instructions. Methods for extracting words (names of ingredients, names of processes, names of utensils) in the extraction process include, but are not limited to, a method using a language vector similarity index and a search process for dictionary data.

[0040] In addition, the names of ingredients included in the cooking instructions may not be consistent, or the names of ingredients included in the cooking instructions may not match the names in the ingredients list included in the cooking recipe. For example, the ingredients list may say "minced pork," but the cooking instructions may say "pork." To determine the identity of ingredients expressed differently in this way, for example, language vectorization technology and language vector similarity indexes provided in machine learning libraries for natural language processing may be used. The standardization unit 13 then extracts the names of ingredients included in the cooking instructions in a form in which the same ingredient name is assigned to ingredients determined to be the same.

[0041] The method of extracting numerical information (e.g., information on the amount of an ingredient, duration-related phrases, etc.) in the extraction process performed by the standardization unit 13 is to use a regular expression processing library compatible with various programming languages ​​to recognize words such as "g" and "min" that represent units of amount or duration, and to select the numerical value associated with the word, thereby making it possible to extract the amount of an ingredient such as "100g" and duration-related phrases such as "10 minutes."

[0042] Next, in the standardization process, the standardization unit 13 standardizes the names of ingredients extracted from the cooking instructions to standard ingredient names by referring to an ingredient name table in which a predetermined number of standard ingredient names are listed, standardizes the names of processes extracted from the cooking instructions to standard process names by referring to a processing name table in which a predetermined number of standard processing names are listed, and standardizes the amounts of ingredients extracted from the cooking recipe to standard ingredient amounts by referring to an ingredient amount table in which a predetermined number of standard ingredient amounts are listed.

[0043] In addition, when the cooking instruction includes the name of an utensil used to perform the processing, the standardization unit 13 refers to an utensil name table in which a predetermined number of standard utensil names are listed, and standardizes the name of the utensil included in the cooking instruction identified by the cooking instruction identification unit 12 to the standard utensil name.

[0044] The above-mentioned material name table, process name table, material amount table, and tool name table are stored in the table storage unit 22 of the storage device 20.

[0045] Then, the required time prediction unit 14 uses the results of machine learning on the accumulated training data to determine a predicted required time to perform the cooking specified in the cooking instruction using ingredients with standard ingredient names in standard ingredient amounts standardized by the standardization unit 13, and processing with standard process names.

[0046] In addition, when the cooking instruction includes the name of an utensil used to perform the processing, the required time prediction unit 14 uses the results of machine learning execution of the accumulated training data to determine a predicted required time to perform the cooking specified in the cooking instruction by processing with a standard processing name using ingredients with standard ingredient names in standard ingredient amounts standardized by the standardization unit 13 and utensils with standard utensil names.

[0047] Furthermore, when the cooking instruction includes a time period related phrase related to the processing time required for processing, the required time prediction unit 14 determines the predicted required time taking into account the standard processing time.

[0048] 2 is a diagram for explaining the steps of the cooking time prediction method performed by the cooking time prediction device 10. As shown in the figure, the cooking recipe acquisition unit 11 acquires a cooking recipe including cooking instructions and an ingredients list (cooking recipe acquisition step). In this example, the cooking recipe includes three cooking instructions: "(1) Cut a," "(2) Fry a in "1," and "(3) Boil a in "2." The cooking recipe also includes an ingredients list of "3 pieces of a."

[0049] Next, the cooking instruction identification unit 12 identifies three cooking instructions included in the cooking recipe: “(1) Cut a.”, “(2) Fry a in “1.”, and “(3) Boil a in “2.”” (cooking instruction identification process).

[0050] Next, for a cooking instruction such as "Cut a," the standardization unit 13 standardizes the ingredient name "a" included in the cooking instruction as the standard ingredient name "A" included in the ingredient name table, standardizes the process name "cut" included in the cooking instruction as the standard process name "cut" included in the process name table, and standardizes the ingredient amount of "a" as the standard ingredient amount of "3 pieces" (standardization process). In addition, for a cooking instruction such as "Stir-fry the a in "1", the standardization unit 13 standardizes the ingredient name "a" included in the cooking instruction as the standard ingredient name "A" included in the ingredient name table, standardizes the process name "stir-fry" included in the cooking instruction as the standard process name "heat" included in the process name table, and standardizes the ingredient amount of "a" as the standard ingredient amount of "3 pieces" (standardization process). Furthermore, for a cooking instruction such as "Boil "2" a," the standardization unit 13 standardizes the ingredient name "a" included in the cooking instruction as the standard ingredient name "A" included in the ingredient name table, standardizes the process name "boil" included in the cooking instruction as the standard process name "boil" included in the process name table, and standardizes the ingredient amount of "a" as the standard ingredient amount of "3 pieces" (standardization process).

[0051] Next, using the method described below, the required time prediction unit 14 uses the results of machine learning execution of the accumulated training data to determine a predicted required time of "xx minutes" for the cooking instruction sentence "Cut a" (required time prediction process). In addition, the required time prediction unit 14 uses the results of machine learning execution of the accumulated training data to determine a predicted required time of "△△ minutes" for the cooking instruction sentence "Stir-fry a in "1"" (required time prediction process). Furthermore, the required time prediction unit 14 uses the results of executing machine learning on the accumulated training data to determine a predicted required time of "□□ minutes" for the cooking instruction sentence "Boil "2" a." (required time prediction step).

[0052] FIG. 3 is a diagram showing an example of an ingredient name table. As shown in the figure, the ingredient name table lists a number of predetermined standard ingredient names, such as "thinly sliced ​​meat" and "block of meat". In the standardization process, the standardization unit 13 determines which item in the ingredient name table shown in FIG. 3 matches the ingredient name included in the cooking instruction extracted in the extraction process. For example, when the ingredient name extracted in the extraction process is "potato", the standardization unit 13 determines that it matches the standard ingredient name "vegetables_root" indicating root vegetables. Here, the above-mentioned language vector similarity index may be used to determine whether the extracted ingredient name matches each standard ingredient name, or a comparison table showing which item in the ingredient name table each ingredient corresponds to may be prepared and used to make the determination.

[0053] Then, if the extracted ingredient name matches the standard ingredient name, the variable (X1 to X16) of the standard ingredient name is assigned "1", and if the extracted ingredient name does not match the standard ingredient name, the variable (X1 to X16) of the standard ingredient name is assigned "0". In the example shown in FIG. 3, the variable X8 of the standard ingredient name "vegetable_root" indicating a root vegetable has a value of "1", and the variables X1 to X7 and X9 to X16 of the other standard ingredient names have a value of "0". In other words, the standardization unit 13 standardizes the ingredient name "potato" to "vegetable_root". In this way, in this embodiment, the ingredient name is converted into a numerical value to be assigned to the variables X1 to X16. In this embodiment, 16 types of standard material names are listed, but the number can be changed as appropriate.

[0054] FIG. 4 is a diagram showing an example of an ingredient amount table. As shown in the figure, the ingredient amount table lists a number of predetermined standard ingredient amounts, such as "meat amount_meat_g" indicating the number of grams of an ingredient and "ingredients amount_meat_b" indicating the number of chunks of an ingredient. In the standardization process, the standardization unit 13 determines which item in the ingredient amount table shown in FIG. 4 the ingredient amount included in the cooking recipe extracted in the extraction process matches. For example, when the ingredient amount extracted in the extraction process is "3 potatoes", the standardization unit 13 determines that the ingredient amount matches the standard ingredient amount "ingredients amount_vegetables_b" indicating the number of vegetable chunks. Here, the aforementioned language vector similarity index may be used to determine whether the amount of an extracted ingredient matches each standard ingredient amount, or a comparison table may be prepared to indicate which item in the ingredient amount table each ingredient amount corresponds to. For example, when the ingredient amount is described in a cooking recipe in units of L (liters) or in units of cups, the comparison table may be used to convert the ingredient amount into an item included in the ingredient amount table.

[0055] If the amount of the extracted ingredient matches the standard ingredient amount, a numerical value in that unit is assigned to the standard ingredient amount variable (X17 to X24), and if the amount of the extracted ingredient does not match the standard ingredient amount, "0" is assigned to the standard ingredient amount variable (X17 to X24). In the example shown in FIG. 4, the number of the standard ingredient amount "ingredient amount_vegetable_b" indicating the number of vegetable chunks is "3", and the number of other standard ingredient amounts is "0". In other words, the standardization unit 13 standardizes the ingredient amount of "3 potatoes" to the number of the standard ingredient amount "ingredient amount_vegetable_b", which is "3". In this way, in this embodiment, the ingredient amount is converted to a numerical value assigned to the variables X17 to X24. In this embodiment, eight types of standard material amounts are listed, but the number can be changed as appropriate.

[0056] FIG. 5 is a diagram showing an example of a process name table. As shown in the figure, the process name table lists a number of predetermined standard process names such as "cut" and "peel". In the standardization process, the standardization unit 13 determines which item in the process name table shown in FIG. 5 matches the name of the process included in the cooking instruction extracted in the extraction process. For example, when the name of the process extracted in the extraction process is "heat", the standardization unit 13 determines that it matches the standard process name of "heat". Here, the match between the extracted process name and each standard process name may be determined using the above-mentioned language vector similarity index, or a comparison table showing which item in the process name table each process corresponds to may be prepared.

[0057] Then, if the extracted process name matches the standard process name, "1" is assigned to the variable (X25 to X38) of the standard process name, and if the extracted process name does not match the standard process name, "0" is assigned to the variable (X25 to X38) of the standard process name. In the example shown in FIG. 5, the number of standard process names "heating" is "1", and the number of other standard process names is "0". In other words, the standardization unit 13 standardizes the process name "heating" to "heating". In this way, in this embodiment, the process name is converted into a numerical value to be assigned to the variables X25 to X38. In this embodiment, 14 types of standard process names are listed, but the number can be changed as appropriate.

[0058] FIG. 6 is a diagram showing an example of an appliance name table. As shown in the figure, the appliance name table lists a number of predetermined standard appliance names such as "pot" and "frying pan". In the standardization process, the standardization unit 13 determines which item in the appliance name table shown in FIG. 6 the name of the appliance included in the cooking instruction extracted in the extraction process matches. For example, when the name of the appliance extracted in the extraction process is "pot", the standardization unit 13 determines that it matches the standard appliance name "pot". Here, the match between the extracted appliance name and each standard appliance name may be determined using the above-mentioned language vector similarity index, or a comparison table showing which item in the appliance name table each appliance corresponds to may be prepared.

[0059] Then, if the extracted instrument name matches the standard instrument name, a "1" is assigned to the standard instrument name variable (X39 to X45), and if the extracted instrument name does not match the standard instrument name, a "0" is assigned to the standard instrument name variable (X39 to X45). In the example shown in FIG. 6, the number of all standard instrument names is "0". In this way, in this embodiment, the instrument name is converted to a numerical value assigned to the variables X39 to X45. In this embodiment, seven types of standard instrument names are listed, but the number can be changed as appropriate.

[0060] If the ingredients, processes, or tools included in the cooking instructions are not registered in the above-mentioned ingredient name table, process name table, or tool name table, the cooking time for the cooking instructions may not be predicted or the prediction accuracy may be reduced. Therefore, when there is a cooking instruction that includes ingredients, processes, or tools that are not registered in the ingredient name table, process name table, or tool name table, it is preferable to add the time required for cooking determined by a human (correct answer value) to the cooking instructions and use it as additional learning data. In this way, the prediction accuracy and versatility of the time required prediction unit 14 can be improved.

[0061] Fig. 7 is a diagram showing an example of a period table. As shown in the figure, the period table lists an item of "standard processing period_minutes." In the standardization process, the standardization unit 13 determines the period-related words and phrases included in the cooking instructions extracted in the extraction process in units of standard processing periods (i.e., minutes) shown in the period table shown in Fig. 7.

[0062] If the cooking instruction contains a period-related phrase, a number in that unit is assigned to the standard processing period unit variable (X46), and if the cooking instruction does not contain a period-related phrase, "0" is assigned to the standard processing period variable (X46). In the example shown in FIG. 7, the standard processing period is "0". If the period-related phrase contains a phrase such as "10 minutes", "standard processing period_minutes" becomes "10", and if the period-related phrase contains a phrase such as "1 hour", "standard processing period_minutes" becomes "60". In this way, in this embodiment, the period-related phrase is converted into a numerical value to be assigned to the variable X46.

[0063] Next, the time required prediction unit 14 uses the results of machine learning execution of the accumulated training data to determine a predicted time required to perform the cooking specified in the cooking instruction by a process with a standard process name using ingredients with standard ingredient names in standard ingredient amounts standardized by the standardization unit 13 and utensils with standard utensil names. In addition, the time required prediction unit 14 determines the predicted time required in consideration of the standard processing period. Specifically, in this embodiment, the time required prediction unit 14 determines a predicted time required to perform the cooking specified in the cooking instruction by using a linear regression model shown in the following formula 1.

[0064] [Formula 1] Estimated time required = a1·X1+a2·X2+a3·X3+···+a46·X46

[0065] In formula 1, the predicted required time is determined by the sum of the products of the coefficients a1 to a46 and the numerical values ​​such as "0" and "1" assigned to the variables X1 to X46 shown in Figures 3 to 7. Each of these coefficients a1 to a46 is a value determined by machine learning.

[0066] The required time prediction unit 14 determines the predicted required time using a linear regression model of a machine learning library, and examples of the machine learning library may be any library such as LightGBM or Catboost, but are not limited to these. In addition, the more data items to be handled when constructing a model by machine learning, the higher the accuracy, and therefore it is preferable.

[0067] As described above, in this embodiment, the estimated required time is not determined by a human being, but the required time prediction unit 14 determines the estimated required time for performing the cooking specified in the cooking instruction using the results of machine learning of the accumulated training data. As a result, it is possible to obtain a consistent result for the estimated required time in a short time.

[0068] The predicted required time calculated by the required time prediction unit 14 includes a value after the decimal point in most cases. When the predicted required time is displayed on a screen or the like, the value after any decimal point may be rounded off or rounded to an integer in order to make the predicted required time easier to see, but in cases where the predicted required time is used to mechanically combine the steps of each cooking instruction, the predicted required time may be used as is without processing.

[0069] Next, the machine learning process will be described. The cooking recipe acquisition unit 11 acquires one of a plurality of cooking recipes stored in the cooking recipe storage unit 21. This cooking recipe includes a cooking instruction including the names of ingredients used in the cooking and the names of processes related to cooking the ingredients, as well as information on the amounts of the ingredients.

[0070] The cooking instruction identification unit 12 identifies each of the multiple cooking instructions included in the cooking recipe. For example, the cooking instruction identification unit 12 divides the cooking recipe acquired by the cooking recipe acquisition unit 11 into multiple cooking instructions using punctuation marks, with a combination of a subject and a predicate as a basic unit.

[0071] The standardization unit 13 standardizes the names of ingredients included in the cooking instructions to standard ingredient names by referring to an ingredient name table in which a plurality of standard ingredient names are listed, standardizes the names of processes included in the cooking instructions to standard process names by referring to a process name table in which a plurality of standard process names are listed, and standardizes the amounts of ingredients included in the cooking recipe to standard ingredient amounts by referring to an ingredient amount table in which a plurality of standard ingredient amounts are listed.

[0072] In addition, if the cooking instruction includes the name of an utensil used to perform the processing, the standardization unit 13 refers to an utensil name table in which multiple standard utensil names are listed, and standardizes the name of the utensil included in the cooking instruction to one of the standard utensil names.

[0073] Furthermore, when the cooking instruction includes a period-related term related to the processing period for performing processing, the standardization unit 13 standardizes the period-related term included in the cooking instruction to a predetermined standard processing period.

[0074] Next, the required time prediction unit 14 stores training data consisting of a combination of a standardized cooking instruction and information on the time required to cook the food specified in the cooking instruction in the training data storage unit 23, and executes machine learning of the training data. For example, one standardized cooking instruction is labeled with the time required for cooking determined by a human (i.e., a correct answer value), thereby creating one training data. The training data storage unit 23 stores a plurality of training data created in this manner.

[0075] In this embodiment, the required time prediction unit 14 adjusts each coefficient of the linear regression model shown in Equation 1 (coefficients a1, a2, ..., etc. in Equation 1 described above) so that the predicted required time derived by Equation 1 approaches the correct value, and obtains it as the execution result of machine learning.

[0076] Next, the effects obtained by machine learning in this embodiment will be described with a concrete example. First, 500 recipes (cooking instructions and ingredient lists) were prepared, and a total of 3315 cooking instructions contained in those recipes were standardized using a computer program to determine standard ingredient names, standard ingredient amounts, standard cooking names, standard utensil names, and standard processing periods, and machine learning data was created using these as the values ​​of X1 to X46 in formula 1. Then, a linear regression prediction model was constructed (i.e., each coefficient a1 to a46 was derived) using formula 1, to which a total of 2655 cooking instructions contained in 398 of the 500 cooking recipes and correct answers entered by a human were applied, as training data.

[0077] Among the 500 cooking recipes mentioned above, the remaining 102 cooking recipes that were not used as training data were used to predict the time required to cook the cooking instructions (predicted cooking time) using the linear regression prediction model (Formula 1) mentioned above based on the standard ingredient names, standard ingredient amounts, standard recipe names, standard utensil names, and standard processing times for a total of 660 cooking instructions. Figure 8 is a graph showing the relationship between the predicted cooking time and the correct cooking time for recipes with unknown cooking times in this embodiment. As shown in Figure 8, the approximation line showing the relationship between the predicted cooking time: x and the actual cooking time (correct value): y is y = 0.991x, and it was confirmed that a highly accurate linear regression prediction model was obtained.

[0078] <Another embodiment> In the above embodiment, a specific example of the configuration of the cooking time prediction device 10 has been described, but the configuration can be changed as appropriate.

[0079] The configurations disclosed in the above embodiments (including other embodiments, the same applies below) can be applied in combination with configurations disclosed in other embodiments, provided that no contradiction arises. Furthermore, the embodiments disclosed in this specification are illustrative, and the embodiments of the present invention are not limited thereto, and can be appropriately modified within the scope that does not deviate from the purpose of the present invention. [Industrial Applicability]

[0080] INDUSTRIAL APPLICABILITY The present invention can be used in a cooking time prediction device that can quickly and accurately predict the time required to perform a cooking process specified in a cooking instruction. [Explanation of symbols]

[0081] 10: Cooking time prediction device 11: Cooking recipe acquisition section 12: Cooking instructions specification section 13: Standardization Department 14: Time required prediction section 20: Storage device 21: Cooking recipe memory section 22: Table storage section 23: Training data storage unit

Claims

1. a standardization unit for a cooking recipe including information on names of ingredients used in cooking, names of processes related to the cooking of the ingredients, and amounts of the ingredients used in the cooking, standardizing the names of the ingredients to one of the standard ingredient names by referring to an ingredient name table in which a predetermined number of standard ingredient names are listed, standardizing the names of the processes to one of the standard process names by referring to a process name table in which a predetermined number of standard process names are listed, and standardizing the amounts of the ingredients to one of the standard ingredient amounts by referring to an ingredient amount table in which a predetermined number of standard ingredient amounts are listed; and a required time prediction unit that determines a required time predicted to be required when the cooking specified in the cooking instruction included in the cooking recipe is performed by using ingredients of the standard ingredient names in the standard ingredient amounts standardized by the standardization unit and by processing with the standard process names, using the results of machine learning execution of accumulated training data.

2. The cooking instructions include the name of the tool to be used to perform the process, The standardization unit refers to a device name table in which a plurality of standard device names are listed, and standardizes the name of the device included in the cooking instruction to one of the standard device names; 2. The cooking time prediction device according to claim 1, wherein the required time prediction unit uses a result of machine learning execution of accumulated training data to determine the predicted required time predicted to be required when the cooking specified in the cooking instruction is performed by using ingredients having the standard ingredient names in the standard ingredient amounts standardized by the standardization unit and by using utensils having the standard utensil names and processing with the standard process names.

3. The cooking instruction includes a period-related phrase related to a processing period during which the processing is performed, The standardization unit standardizes the period-related phrase included in the cooking instruction to a predetermined standard processing period, The cooking time prediction device according to claim 1 , wherein the required time prediction unit determines the predicted required time taking into account the standard processing period.

4. The cooking time prediction device according to claim 1 , further comprising a cooking instruction identification unit configured to identify each of the plurality of cooking instructions included in the cooking recipe.

Images