Intelligent recipe generation system and method based on artificial intelligence large model
Through artificial intelligence big model, analyzing the dishes scores and elements composition, automatically generating recipes, solving the problem that traditional recipe generation is difficult to meet diversified nutritional needs and control costs, realizing intelligent and personalized recipe generation, improving catering service efficiency and quality.
Patent Information
- Application Number
- CN202411307237.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-19
- Publication Date
- 2025-01-03
AI Technical Summary
Traditional recipe generation relies on manual experience and chefs’ personal preferences, and it is difficult to meet diverse and personalized nutritional needs while controlling costs, resulting in poor dining experience for diners.
Using an intelligent recipe generation system based on artificial intelligence big model, the dishes historical score records are obtained through the interactive all-in-one system, the comprehensive scores are analyzed, the popular dishes are determined, the elements composition and proportion are obtained, and the recipe generation is digitalized, and the Transformer architecture is used to train the recipe generation. Model to realize the automatic generation of recipes.
It has achieved a highly automated and intelligent recipe generation, accurately matched personalized nutritional needs, promoted healthy diet, effectively controlled costs, improved the operating efficiency of catering companies, and adapted to market changes and user preferences.
Smart Images
Figure CN119249267A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an intelligent recipe generation system and method based on an artificial intelligence big model. Background Art
[0002] With the development of artificial intelligence technology, especially breakthroughs in natural language processing and deep learning, large models have become an important tool for handling complex tasks; In a company, different departments are grouped together according to their respective functions and needs. The catering arrangements and food standards between different departments may be different. Employees in the same department eat in the same canteen and eat the same food standards. Therefore, different departments within the same company have formed their own characteristics and standards in terms of catering. In the catering industry, traditional recipe generation relies on manual experience and the personal preferences of chefs. It is difficult to meet diverse and personalized nutritional needs while controlling costs, resulting in diners not having a good dining experience during the meal. Therefore, it is particularly important to develop an intelligent recipe generation and optimization system that can combine large model capabilities and automatically consider multiple factors such as ingredient costs, nutritional balance, and taste preferences. Summary of the invention
[0003] The purpose of the present invention is to provide an intelligent recipe generation system and method based on an artificial intelligence big model to solve the problems raised in the prior art.
[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent recipe method based on an artificial intelligence big model, the method comprising: Step S100: After dining in the cafeteria, the user reviews the dishes through the interactive all-in-one machine system, and obtains the historical rating records of the dishes according to the interactive all-in-one machine system; Step S200: Analyze the comprehensive scores of any dishes according to historical score records, confirm popular dishes, and obtain a set of popular dishes; Step S300: in the popular dish set, obtain the element composition and element proportion of the dish, form an element composition set of any dish, and divide the dish categories; Step S400: digitally processing popular dishes according to the element composition set of step S300; Step S500: training the large model according to the digital processing result of step S400 to obtain a recipe generation model; Step S600: Automatically generate recipes based on the recipe generation model.
[0005] Further, in step S100, in the interactive all-in-one machine system, historical rating records of dishes by users are obtained, and historical rating records of the same dish are summarized to obtain a historical rating record set A={a1, a2, a3, ..., an} of any dish, wherein a1, a2, a3, ..., an represent the 1st, 2nd, 3rd, ..., nth historical rating records of any dish, respectively; Where an={an1, an2, an3, ..., anm}, where an1, an2, an3, ..., anm respectively represent the specific scores of the 1st, 2nd, 3rd, ..., mth indicators in the nth historical rating record of any dish; In a company, different departments are grouped together according to their respective functions and needs. For example, employees in the R&D department may eat in the same canteen, with the same food standards, and can enter and enjoy the same meals by swiping their employee cards. At the same time, workers in the production department will also eat in their own exclusive canteen, and although the food standards are similar, they may vary according to the nature of the work; However, the catering arrangements and food standards between different departments may vary. The R&D department may provide more abundant healthy meal options to support employees' mental work, while the production department may focus more on high-energy, high-protein meals to meet the needs of manual workers. In this way, different departments within the same company have formed their own characteristics and standards in terms of catering; In order to provide clearer feedback on users' ratings of dishes, the canteen will use an interactive all-in-one machine to allow users to review dishes after they have finished their meals. There are multiple indicators for rating dishes, including color, taste, sweetness, etc. Multiple rating indicators can more accurately reflect the ratings of dishes.
[0006] Further, step S200 includes: Step S201: In the historical scoring record set of any dish, the specific score of each indicator in the nth historical scoring record is calculated according to the formula:
[0007] Among them, X in It is represented by the historical comprehensive score of the nth historical score record of the ith dish, X inr It is represented by the specific score of the rth indicator in the nth historical score record of the ith dish, a ir Expressed as the weight of the rth indicator of the ith dish, the historical comprehensive score record set X={x1, x2, x3, ..., xn} of any dish is obtained, where x1, x2, x3, ..., xn represent the historical comprehensive scores of the 1st, 2nd, 3rd, ..., nth historical score records of any dish respectively; Step S202: Summarize the historical comprehensive scores of any dish according to the formula:
[0008] Among them, X i综 It is represented as the comprehensive score of the i-th dish, X if It is represented as the historical comprehensive score of the fth historical score record of the ith dish, and the comprehensive score of any dish is obtained; Step S203: when the comprehensive score of a dish is greater than or equal to the comprehensive score threshold, the dish is confirmed to be a popular dish, and the popular dishes are aggregated to obtain a popular dish set and a popular dish comprehensive score set; Because the importance of different indicators in the scoring index of dishes is different, it is necessary to calculate the comprehensive score of the dishes, and set the dishes whose comprehensive scores meet the standards as popular dishes, so as to improve the accuracy of judging them as popular dishes.
[0009] Furthermore, step S300 includes: Step S301: confirming the elements of a popular dish in the popular dish set to obtain a set of elements of the popular dish; Step S302: In the element composition set of a popular dish, the weight of a certain element in the popular dish is confirmed, and the proportion of a certain element in the popular dish is calculated according to the following formula:
[0010] Among them, Y ez It is expressed as the proportion of the zth element in the eth most popular dish, A ez A is the weight of the zth element in the eth most popular dish. eb It is represented as the weight of the bth element in the eth popular dish, C is represented as the number of elements in the eth popular dish, and the element ratio set of a popular dish is obtained; Step S303: In the element proportion set of the e-th popular dish, when the proportion of a certain element is greater than or equal to the proportion threshold, the element is set as a characteristic element, and the characteristic element set of the e-th popular dish is obtained as J e ; Step S304: Obtaining a set of characteristic elements T of different types of dishes m , m represents the dishes of type m; set J e With the set T m For comparison, when there is When , the category of the e-th popular dish is determined to be m, and a set of popular dishes of the same category is obtained; Because the types of dishes are divided into major meat dishes, semi-meat dishes, minor meat dishes, vegetarian dishes, etc., and each type of dish has different elements, it is necessary to confirm what the characteristic elements are in the dishes, and judge based on the characteristic elements of the dishes to determine which category the dish belongs to.
[0011] Because the types of dishes are divided into major meat dishes, semi-meat dishes, minor meat dishes, vegetarian dishes, etc., and each type of dish has different elements, it is necessary to confirm what the characteristic elements are in the dishes, and judge based on the characteristic elements of the dishes to determine which category the dish belongs to.
[0012] Furthermore, step S400 includes: Step S401: In a set of popular dishes in a certain dish category, the number of comments and the number of purchases of a certain popular dish are collected, and the number of comments and the number of purchases of all popular dishes are summarized to obtain a set of the number of comments and the set of the number of purchases of a certain dish category; Step S402: Input the comprehensive score set, the number of reviews set and the number of purchases set of a certain dish category into the normalization formula respectively. According to the normalization formula:
[0013] Where L represents the normalized value, w represents any unnormalized value in the set, and w min Expressed as the minimum value of the set, w max denoted as the maximum value of the set, d as the weight of the normalization formula, and the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category are obtained; Step S403: In the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category, obtain the normalized comprehensive score of the e-th popular dish as Q e 、The number of reviews is P e 、The number of purchases is S e , according to the digital formula:
[0014] Among them, K e It is represented by the popularity of the e-th popular dish, t 1 ,t 2 ,t 3 They are respectively expressed as the weight of the comprehensive score, the weight of the number of reviews, and the weight of the number of purchases; the popularity of all popular dishes in a certain dish category is summarized to obtain the popularity set of all dishes in a certain dish category; Because in the historical rating records, the number of reviews and the number of purchases of a dish may be different, which will affect the judgment of the popularity of the dish. Therefore, it is necessary to normalize the comprehensive score, the number of reviews and the number of purchases to improve the accuracy of calculating popularity.
[0015] Furthermore, step S500 includes: Step S501: in a popularity set of a certain dish category, obtain the nutrient value and cost of a certain popular dish, and input the popularity, nutrient value and cost of the certain popular dish into the training large model of the Transformer architecture; Step S502: input the popularity, nutrient value and cost of popular dishes in all dish categories into the training large model, perform training, and obtain a recipe generation model; The input data of dish cost, nutrient value, dish type, and popularity are structured into a format that the model can process. The encoder is responsible for converting the input features into a high-dimensional representation. The decoder gradually generates a complete recipe based on the encoder output and the generated partial sequence. The loss function of sequence generation is used to measure the difference between the sequence generated by the model and the actual sequence: during training, given a partial sequence, the model predicts the next component or step. The difference is calculated based on the probability distribution predicted by the model and the actual target distribution, and is used for backpropagation to adjust the model parameters. The Transformer model is trained through autoregression, that is, each time the partial sequence generated previously is used to predict the next element.
[0016] Furthermore, in step S600, the real-time cost budget and real-time nutrient requirement value of the canteen are obtained, and the real-time cost budget and real-time nutrient requirement value are input into the recipe generation model to automatically generate a recipe; Recipes are automatically generated based on real-time cost budgets and real-time nutrient requirements, avoiding reliance on manual experience and the chef's personal preferences, meeting diverse and personalized nutritional needs while controlling costs.
[0017] In order to better implement the above method, an intelligent recipe generation system based on an artificial intelligence large model is also proposed. The system includes a historical data module, a comprehensive scoring module, a dish category module, a digital processing module, a model building module and a real-time generation module; Historical data module: After dining in the cafeteria, users can review dishes through the interactive all-in-one machine system and obtain historical rating records of dishes based on the interactive all-in-one machine system; Comprehensive scoring module: Analyze the comprehensive scores of any dishes based on historical scoring records, identify popular dishes, and obtain a collection of popular dishes; Dish category module: In the popular dish set, obtain the element composition and element proportion of the dish, form the element composition set of any dish, and divide the dish category; Digital processing module: digitally processing the popular dishes according to the element composition set of step S300; Model building module: training the large model according to the digital processing result of step S400 to obtain a recipe generation model; Real-time generation module: Automatically generates recipes based on the recipe generation model.
[0018] Furthermore, the dish category module includes a dish element proportion calculation unit and a dish category determination unit; Dish element ratio calculation unit: confirm the elements of popular dishes to obtain the element composition set of popular dishes; calculate the ratio of each element in the popular dishes according to the element composition set of popular dishes to obtain the element ratio set of popular dishes; Unit for determining dish category: when the element proportion is greater than or equal to the proportion threshold, the element is set as a characteristic element to obtain a characteristic element set of popular dishes; characteristic element sets of different types of dishes are obtained, and the characteristic element set of popular dishes is compared with the characteristic element sets of different types of dishes to obtain a set of popular dishes of the same type.
[0019] Further, the digital processing module includes a normalization processing unit and a popularity calculation unit; Normalization processing unit: in the set of popular dishes in the dish category, the number of comments and the number of purchases of the popular dishes are collected, and the number of comments and the number of purchases of all popular dishes are summarized to obtain the set of the number of comments and the set of the number of purchases of the dish category; the comprehensive score set, the set of the number of comments and the set of the number of purchases of the dish category are respectively input into the normalization formula to obtain the normalized comprehensive score set, the set of the number of comments and the set of the number of purchases of the dish category; Popularity calculation unit: in the normalized comprehensive score set, review number set and purchase number set of the dish category, obtain the normalized comprehensive score, review number and purchase number of the popular dishes, and calculate the popularity of the popular dishes; summarize the popularity of all popular dishes in the dish category to obtain the popularity set of all dishes in the dish category.
[0020] Compared with the prior art, the beneficial effects of the present invention are: commenting on dishes through the interactive all-in-one system, obtaining the historical rating records of dishes according to the interactive all-in-one system; analyzing the comprehensive ratings of any dish, confirming the popular dishes, and obtaining a set of popular dishes; obtaining the element composition and element proportion of the dish, forming an element composition set of any dish, and dividing the types of dishes; digitally processing the popular dishes according to the element composition set; training the large model according to the digital processing results to obtain a recipe generation model; and automatically generating recipes according to the recipe generation model. It can achieve high automation and intelligence in recipe generation, significantly improve the efficiency and quality of catering services; accurately match personalized nutritional needs and promote healthy eating; effectively control costs and improve the operating efficiency of catering companies; continuously learn and optimize to adapt to market changes and dynamic adjustments of user preferences. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1 The present invention is a flowchart of an intelligent recipe generation method based on an artificial intelligence large model. DETAILED DESCRIPTION
[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0023] See also Figure 1 The present invention provides a technical solution: an intelligent recipe generation method based on an artificial intelligence big model, the method comprising: Step S100: After dining in the cafeteria, the user reviews the dishes through the interactive all-in-one machine system, and obtains the historical rating records of the dishes according to the interactive all-in-one machine system; In step S100, in the interactive all-in-one machine system, historical rating records of dishes by users are obtained, and historical rating records of the same dish are summarized to obtain a historical rating record set A={a1, a2, a3, ..., an} of any dish, wherein a1, a2, a3, ..., an represent the 1st, 2nd, 3rd, ..., nth historical rating records of any dish, respectively; Where an={an1, an2, an3, ..., anm}, where an1, an2, an3, ..., anm respectively represent the specific scores of the 1st, 2nd, 3rd, ..., mth indicators in the nth historical rating record of any dish; For example, the set of historical rating records for the first dish is {the first historical rating record, the second historical rating record, the third historical rating record, the fourth historical rating record}; The indicator set in the first historical rating record of the first dish is {the specific rating of the first indicator is 4.5, the specific rating of the second indicator is 4.8, and the specific rating of the third indicator is 4.7}.
[0024] Step S200: Analyze the comprehensive scores of any dishes according to historical score records, confirm popular dishes, and obtain a set of popular dishes; Wherein, step S200 includes: Step S201: In the historical scoring record set of any dish, the specific score of each indicator in the nth historical scoring record is calculated according to the formula:
[0025] Among them, X in It is represented by the historical comprehensive score of the nth historical score record of the ith dish, X inr It is represented by the specific score of the rth indicator in the nth historical score record of the ith dish, a ir Expressed as the weight of the rth indicator of the ith dish, the historical comprehensive score record set X={x1, x2, x3, ..., xn} of any dish is obtained, where x1, x2, x3, ..., xn represent the historical comprehensive scores of the 1st, 2nd, 3rd, ..., nth historical score records of any dish respectively; Step S202: Summarize the historical comprehensive scores of any dish according to the formula:
[0026] Among them, X i综 It is represented as the comprehensive score of the i-th dish, X if It is represented as the historical comprehensive score of the fth historical score record of the ith dish, and the comprehensive score of any dish is obtained; Step S203: when the comprehensive score of a dish is greater than or equal to the comprehensive score threshold, the dish is confirmed to be a popular dish, and the popular dishes are aggregated to obtain a popular dish set and a popular dish comprehensive score set; For example, if the weight of the first indicator is 0.4, the weight of the second indicator is 0.25, and the weight of the third indicator is 0.3, then the historical comprehensive score of the first historical score record of the first dish is 4.41; The historical comprehensive score of the second historical rating record of the first dish is 4.68, and the historical comprehensive score of the third historical rating record of the first dish is 4.34. The calculated comprehensive score of the first dish is 4.48, and the comprehensive score threshold is 4.35, so the first dish is a popular dish.
[0027] Step S300: in the popular dish set, obtain the element composition and element proportion of the dish, form an element composition set of any dish, and divide the dish categories; Wherein, step S300 includes: Step S301: confirming the elements of a popular dish in the popular dish set to obtain a set of elements of the popular dish; Step S302: In the element composition set of a popular dish, the weight of a certain element in the popular dish is confirmed, and the proportion of a certain element in the popular dish is calculated according to the following formula:
[0028] Among them, Y ez It is expressed as the proportion of the zth element in the eth most popular dish, A ez A is the weight of the zth element in the eth most popular dish. eb It is represented as the weight of the bth element in the eth popular dish, C is represented as the number of elements in the eth popular dish, and the element ratio set of a popular dish is obtained; Step S303: In the element proportion set of the e-th popular dish, when the proportion of a certain element is greater than or equal to the proportion threshold, the element is set as a characteristic element, and the characteristic element set of the e-th popular dish is obtained as J e ; Step S304: Obtaining a set of characteristic elements T of different types of dishes m , m represents the dishes of type m; set J e With the set T m For comparison, when there is When , the category of the e-th popular dish is determined to be m, and a set of popular dishes of the same category is obtained; For example, the element composition set of the first popular dish is {element a, element b, element c, element d}, the weight of element a in the popular dishes is 400g, the weight of element b in the popular dishes is 150g, the weight of element c in the popular dishes is 50g, and the weight of element d in the popular dishes is 25g. The calculated proportion of element a is 0.64, the proportion of element b is 0.25, the proportion of element c is 0.08, and the proportion of element d is 0.04. The proportion threshold is 0.35, and the characteristic element set of the first popular dish is {element a}; The characteristic element set of the first category of dishes is {element a, element g, element k}. By comparison, the first popular dish is determined to be the first category of dishes.
[0029] Step S400: digitally processing popular dishes according to the element composition set of step S300; Wherein, step S400 includes: Step S401: In a set of popular dishes in a certain dish category, the number of comments and the number of purchases of a certain popular dish are collected, and the number of comments and the number of purchases of all popular dishes are summarized to obtain a set of the number of comments and the set of the number of purchases of a certain dish category; Step S402: Input the comprehensive score set, the number of reviews set and the number of purchases set of a certain dish category into the normalization formula respectively. According to the normalization formula:
[0030] Where L represents the normalized value, w represents any unnormalized value in the set, and w min Expressed as the minimum value of the set, w max denoted as the maximum value of the set, d as the weight of the normalization formula, and the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category are obtained; Step S403: In the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category, obtain the normalized comprehensive score of the e-th popular dish as Q e 、The number of reviews is P e 、The number of purchases is S e , according to the digital formula:
[0031] Among them, K e It is represented by the popularity of the e-th popular dish, t 1 ,t 2 ,t 3They are respectively expressed as the weight of the comprehensive score, the weight of the number of reviews, and the weight of the number of purchases; the popularity of all popular dishes in a certain dish category is summarized to obtain the popularity set of all dishes in a certain dish category; For example, the set of the number of reviews for the first dish category is {the number of reviews for the first popular dish is 1000, the number of reviews for the second popular dish is 1500, the number of reviews for the third popular dish is 2000, and the number of reviews for the fourth popular dish is 2500}; The normalized set of review numbers for the first dish category is {the number of reviews for the first popular dish is 0.067, the number of reviews for the second popular dish is 0.4, the number of reviews for the third popular dish is 0.733, and the number of reviews for the fourth popular dish is 1.067}.
[0032] Step S500: training the large model according to the digital processing result of step S400 to obtain a recipe generation model; Wherein, step S500 includes: Step S501: in a popularity set of a certain dish category, obtain the nutrient value and cost of a certain popular dish, and input the popularity, nutrient value and cost of the certain popular dish into the training large model of the Transformer architecture; Step S502: input the popularity, nutrient value and cost of popular dishes in all dish categories into the training large model, perform training, and obtain a recipe generation model; For example, in the training model of the Transformer architecture, constraints are added that each dish cannot be served more than twice a week; each time a recipe is generated, there must be at least one dish in each dish category.
[0033] Step S600: Automatically generate recipes based on the recipe generation model In step S600, the real-time cost budget and real-time nutrient requirement value of the cafeteria are obtained, and the real-time cost budget and real-time nutrient requirement value are input into the recipe generation model to automatically generate a recipe.
[0034] In order to better implement the above method, an intelligent recipe generation system based on an artificial intelligence large model is also proposed. The system includes a historical data module, a comprehensive scoring module, a dish category module, a digital processing module, a model building module and a real-time generation module; Historical data module: After dining in the cafeteria, users can review dishes through the interactive all-in-one machine system and obtain historical rating records of dishes based on the interactive all-in-one machine system; Comprehensive scoring module: Analyze the comprehensive scores of any dishes based on historical scoring records, identify popular dishes, and obtain a collection of popular dishes; Dish category module: In the popular dish set, obtain the element composition and element proportion of the dish, form the element composition set of any dish, and divide the dish category; Among them, the dish category module includes a dish element proportion calculation unit and a dish category determination unit; Dish element ratio calculation unit: confirm the elements of popular dishes to obtain the element composition set of popular dishes; calculate the ratio of each element in the popular dishes according to the element composition set of popular dishes to obtain the element ratio set of popular dishes; Unit for determining the type of dishes: when the element ratio is greater than or equal to the ratio threshold, the element is set as a characteristic element to obtain a characteristic element set of popular dishes; characteristic element sets of dishes of different types are obtained, and the characteristic element set of popular dishes is compared with the characteristic element sets of dishes of different types to obtain a set of popular dishes of the same type; Digital processing module: digitally processing the popular dishes according to the element composition set of step S300; Wherein, the digital processing module includes a normalization processing unit and a popularity calculation unit; Normalization processing unit: in the set of popular dishes in the dish category, the number of comments and the number of purchases of the popular dishes are collected, and the number of comments and the number of purchases of all popular dishes are summarized to obtain the set of the number of comments and the set of the number of purchases of the dish category; the comprehensive score set, the set of the number of comments and the set of the number of purchases of the dish category are respectively input into the normalization formula to obtain the normalized comprehensive score set, the set of the number of comments and the set of the number of purchases of the dish category; Popularity calculation unit: in the normalized comprehensive score set, review number set and purchase number set of the dish category, obtain the normalized comprehensive score, review number and purchase number of the popular dish, and calculate the popularity of the popular dish; summarize the popularity of all the popular dishes in the dish category to obtain the popularity set of all the dishes in the dish category; Model building module: training the large model according to the digital processing result of step S400 to obtain a recipe generation model; Real-time generation module: Automatically generates recipes based on the recipe generation model.
[0035] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.
Claims
1. An intelligent recipe generation method based on an artificial intelligence big model, characterized by: Step S100: After dining in the cafeteria, the user reviews the dishes through the interactive all-in-one machine system, and obtains the historical rating records of the dishes according to the interactive all-in-one machine system; Step S200: Analyze the comprehensive scores of any dishes according to historical score records, confirm popular dishes, and obtain a set of popular dishes; Step S300: in the popular dish set, obtain the element composition and element proportion of the dish, form an element composition set of any dish, and divide the dish categories; Step S400: digitally processing popular dishes according to the element composition set of step S300; Step S500: According to the digital processing result of step S400, the large model is trained to obtain a recipe generation model; Step S600: Automatically generate recipes based on the recipe generation model.
2. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 1, characterized in that: In step S100, in the interactive all-in-one machine system, historical rating records of dishes by users are obtained, and historical rating records of the same dish are summarized to obtain a historical rating record set A={a1, a2, a3, ..., an} of any dish, where a1, a2, a3, ..., an represent the 1st, 2nd, 3rd, ..., nth historical rating records of any dish, respectively; Among them, an={an1, an2, an3, ..., anm}, where an1, an2, an3, ..., anm respectively represent the specific scores of the 1st, 2nd, 3rd, ..., mth indicators in the nth historical rating record of any dish.
3. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 2 is characterized in that: The step S200 includes the following steps: Step S201: In the historical scoring record set of any dish, the specific score of each indicator in the nth historical scoring record is calculated according to the formula: ； Among them, X in It is represented by the historical comprehensive score of the nth historical score record of the ith dish, X inr It is represented by the specific score of the rth indicator in the nth historical score record of the ith dish, a ir Expressed as the weight of the rth indicator of the ith dish, the historical comprehensive score record set X={x1, x2, x3, ..., xn} of any dish is obtained, where x1, x2, x3, ..., xn represent the historical comprehensive scores of the 1st, 2nd, 3rd, ..., nth historical score records of any dish respectively; Step S202: Summarize the historical comprehensive scores of any dish according to the formula: ； Among them, X i综 It is represented as the comprehensive score of the i-th dish, X if It is represented as the historical comprehensive score of the fth historical score record of the ith dish, and the comprehensive score of any dish is obtained; Step S203: When the comprehensive score of a dish is greater than or equal to the comprehensive score threshold, the dish is confirmed to be a popular dish, and the popular dishes are aggregated to obtain a popular dish set and a comprehensive score set of the popular dishes.
4. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 3 is characterized in that: The step S300 includes the following steps: Step S301: in the popular dish set, confirm the elements of a popular dish to obtain the element composition set of the popular dish; Step S302: In the element composition set of a popular dish, the weight of a certain element in the popular dish is confirmed, and the proportion of a certain element in the popular dish is calculated according to the following formula: ； Among them, Y ez It is expressed as the proportion of the zth element in the eth most popular dish, A ez A is the weight of the zth element in the eth most popular dish. eb It is represented as the weight of the bth element in the eth popular dish, C is represented as the number of elements in the eth popular dish, and the element ratio set of a popular dish is obtained; Step S303: In the element proportion set of the e-th popular dish, when the proportion of a certain element is greater than or equal to the proportion threshold, the element is set as a characteristic element, and the characteristic element set of the e-th popular dish is obtained as J e ; Step S304: Obtaining a set of characteristic elements T of different types of dishes m , m represents the dishes of type m; set J e With the set T m For comparison, when there is , then the category of the e-th popular dish is determined to be m, and a set of popular dishes of the same category is obtained.
5. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 4 is characterized in that: The step S400 includes the following steps: Step S401: In a set of popular dishes in a certain dish category, the number of comments and the number of purchases of a certain popular dish are collected, and the number of comments and the number of purchases of all popular dishes are summarized to obtain a set of the number of comments and the set of the number of purchases of a certain dish category; Step S402: Input the comprehensive score set, the number of reviews set and the number of purchases set of a certain dish category into the normalization formula respectively. According to the normalization formula: ； Where L represents the normalized value, w represents any unnormalized value in the set, and w min Expressed as the minimum value of the set, w max denoted as the maximum value of the set, d as the weight of the normalization formula, and the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category are obtained; Step S403: In the normalized comprehensive score set, review quantity set, and purchase frequency set of a certain dish category, obtain the normalized comprehensive score of the e-th popular dish as Q e 、The number of reviews is P e 、The number of purchases is S e , according to the digital formula: ； Among them, K e It represents the popularity of the e-th popular dish, t1, t2, and t3 represent the weight of the comprehensive score, the weight of the number of reviews, and the weight of the number of purchases, respectively. The popularity of all popular dishes in a certain dish category is summarized to obtain the popularity set of all dishes in a certain dish category.
6. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 5, characterized in that: The step S500 includes the following steps: Step S501: in a popularity set of a certain dish category, obtain the nutrient value and cost of a certain popular dish, and input the popularity, nutrient value and cost of the certain popular dish into the training large model of the Transformer architecture; Step S502: The popularity, nutrient value and cost of popular dishes in all dish categories are input into the training large model for training to obtain a recipe generation model.
7. The method for generating intelligent recipes based on an artificial intelligence big model according to claim 6, characterized in that: In step S600, the real-time cost budget and real-time nutrient requirement value of the cafeteria are obtained, and the real-time cost budget and real-time nutrient requirement value are input into the recipe generation model to automatically generate a recipe.
8. An intelligent recipe generation system based on an artificial intelligence big model, used to implement an intelligent recipe generation method based on an artificial intelligence big model as described in any one of claims 1 to 7, characterized in that: The system includes: a historical data module, a comprehensive scoring module, a dish category module, a digital processing module, a model building module and a real-time generation module; The historical data module: after dining in the cafeteria, users review the dishes through the interactive all-in-one machine system, and obtain the historical rating records of the dishes according to the interactive all-in-one machine system; The comprehensive scoring module: analyzes the comprehensive scores of any dishes according to historical scoring records, identifies popular dishes, and obtains a set of popular dishes; The dish category module: in the popular dish set, obtains the element composition and element proportion of the dish, forms an element composition set of any dish, and divides the dish category; The digital processing module: performs digital processing on the popular dishes according to the element composition set of step S300; The model building module: training the large model according to the digital processing result of step S400 to obtain a recipe generation model; The real-time generation module automatically generates recipes based on the recipe generation model.
9. The intelligent recipe generation system based on artificial intelligence big model according to claim 8 is characterized in that: The dish category module includes a dish element proportion calculation unit and a dish category determination unit; The unit for calculating the proportion of elements of the dishes: confirming the elements of the popular dishes to obtain the element composition set of the popular dishes; calculating the proportion of each element in the popular dishes according to the element composition set of the popular dishes to obtain the element proportion set of the popular dishes; The unit for determining the type of dishes: when the proportion of an element is greater than or equal to a proportion threshold, the element is set as a characteristic element to obtain a characteristic element set of popular dishes; characteristic element sets of different types of dishes are obtained, and the characteristic element set of popular dishes is compared with the characteristic element sets of different types of dishes to obtain a set of popular dishes of the same type.
10. The intelligent recipe generation system based on artificial intelligence big model according to claim 8 is characterized in that: The digital processing module includes a normalization processing unit and a popularity calculation unit; The normalization processing unit: collects the number of comments and the number of purchases of the popular dishes in the set of popular dishes of the dish category, summarizes the number of comments and the number of purchases of all the popular dishes, and obtains the set of the number of comments and the set of the number of purchases of the dish category; inputs the comprehensive score set, the set of the number of comments, and the set of the number of purchases of the dish category into the normalization formula respectively, and obtains the normalized comprehensive score set, the set of the number of comments, and the set of the number of purchases of the dish category; The popularity calculation unit: obtains the normalized comprehensive score, number of comments, and number of purchases of popular dishes from the normalized comprehensive score set, number of comments set, and number of purchases of the dish category, and calculates the popularity of the popular dishes; summarizes the popularity of all popular dishes in the dish category to obtain a popularity set of all dishes in the dish category.