A food formula generation method and system based on a knowledge graph

Abstract

The application discloses a food formula generation method and system based on a knowledge graph, relates to the field of food formula, and mainly has the following scheme: a food material knowledge graph is constructed based on a graph data set tool, an initial formula is input into the food material knowledge graph, substitute food materials of different food materials in the initial formula are acquired, food material data meeting requirements are selected from an initial food material set according to formula improvement requirements, and an improved formula set is generated; food materials are arranged and combined to obtain a plurality of improved formulas; an evaluation index set of each improved formula is calculated, and a comprehensive analysis is performed according to parameters in the evaluation index set to obtain a formula comprehensive evaluation index; and improved formulas meeting standards are selected; the method solves the problem that most of prior art lacks a scientific and systematic comprehensive evaluation method, cannot quickly and efficiently meet special needs of different consumers, cannot realize large-scale personalized food formula customization, and limits the development space of food enterprises in a high-end customization market.

Description

Technical Field

[0001] This invention relates to the field of food formulation, specifically to a method and system for generating food formulations based on knowledge graphs. Background Technology

[0002] Food formulation design plays a crucial role in the development of the food industry. An excellent food formulation not only enhances the taste, flavor, and nutritional value of food, meeting the increasingly diverse and personalized needs of consumers—such as the development of healthy food formulations like low-sugar, low-fat, and high-fiber options—but also helps food companies reduce production costs, improve efficiency, and stand out in fierce market competition. With rising living standards and heightened health awareness, the demand for high-quality, nutritious food continues to grow, expanding the application prospects of food formulation methods and creating greater economic benefits for businesses.

[0003] In the competitive food industry, product homogenization is prevalent. To meet the diverse taste preferences, nutritional needs, and special dietary requirements of consumers, such as those requiring specialized formulas for diabetics or allergies, food companies need to continuously adjust and optimize their formulas to achieve product differentiation and personalization, thereby gaining market share. Furthermore, as consumers become increasingly concerned about food safety and quality, companies also need to adjust their formulas promptly, reducing or avoiding the use of potentially harmful additives and preservatives, and selecting safer, more natural ingredients to adapt to market trends and regulatory requirements. Simultaneously, factors such as raw material supply, price fluctuations, and the emergence of new research findings may also prompt companies to adjust their food formulas to reduce production costs and improve product quality and stability.

[0004] Traditional food formulation development often relies on experience and trial and error. For adjusted food formulations, existing technologies mostly lack a scientific and systematic comprehensive evaluation method, requiring a large number of experiments and adjustments, consuming a lot of time, human and material resources, and with a low success rate.

[0005] With increasingly diverse and personalized consumer demands, existing food formulation methods are struggling to quickly and efficiently meet the specific needs of different consumers, making it impossible to achieve large-scale personalized food formulation customization and limiting the development space of food companies in the high-end customization market. Summary of the Invention

[0006] (a) Technical problems to be solved

[0007] To address the shortcomings of existing technologies, this invention provides a food formula generation method based on knowledge graphs, which at least solves the problem that existing food formula generation methods are unable to quickly and efficiently meet the special needs of different consumers and cannot achieve large-scale personalized food formula customization.

[0008] (II) Technical Solution

[0009] To achieve the above objectives, the present invention provides the following technical solution: a food formula generation method based on knowledge graphs, comprising:

[0010] Step 1: Obtain the ingredient data of the known knowledge graph, and construct the ingredient knowledge graph based on the graph data group tool. Input the initial recipe into the ingredient knowledge graph, obtain the substitute ingredients for different ingredients in the initial recipe, and form the initial set of ingredients for screening.

[0011] Step 2: Determine the recipe improvement requirements, and based on the requirements, select the ingredients that meet the requirements from the initial ingredient set to generate the secondary ingredient set. Then, merge the initial ingredients of the initial recipe with the secondary ingredient set to generate the improved recipe set.

[0012] Step 3: Arrange and combine each ingredient in each data group in the improved recipe set; to obtain several improved recipes;

[0013] Step 4: Based on the ingredient data of various ingredients in different improved recipes, calculate the evaluation index set for each improved recipe, and conduct a comprehensive analysis based on the parameters in the evaluation index set to obtain the comprehensive evaluation index of the recipe.

[0014] Step 5: Following the methods in Steps 1 to 4, calculate the comprehensive evaluation index of the initial formula. Analyze the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula to obtain the deviation index. Based on the deviation index, select the improved formulas that meet the standards, and sort the improved formulas that meet the standards according to the comprehensive evaluation index to generate a formula execution priority chart.

[0015] In the preferred embodiment of the knowledge graph-based food recipe generation method described above, the recipe improvement requirements include cost restrictions and ingredient component restrictions. When performing secondary screening based on the recipe improvement requirements, if the restriction is a cost restriction, the price cost of the initial ingredient is compared with the price cost of the substitute ingredient. If the price cost of the initial ingredient is greater than the price cost of the substitute ingredient, the substitute ingredient meets the requirement, and a secondary screening ingredient set is generated. If the recipe improvement requirement is an ingredient component restriction, which means that the components of some ingredients are restricted, then the substitute ingredients are not allowed to contain the restricted components, and the ingredient data of the substitute ingredients that do not contain the restricted components are used to generate a secondary screening ingredient set.

[0016] In the preferred embodiment of the above-mentioned knowledge graph-based food formula generation method, the secondary screening ingredient set contains different data groups, each data group corresponding to one ingredient in the initial formula, and the ingredients in this data group are all substitute ingredients for the ingredients in the initial formula; when one or more initial ingredients have no substitute ingredients, this initial ingredient is included in the data group of the secondary screening ingredient set.

[0017] In the preferred embodiment of the above-mentioned knowledge graph-based food formula generation method, if the restriction requirement is a cost restriction requirement, the various initial ingredients in the initial formula are added to the data group of the secondary screening ingredient set to form an improved formula set; if the formula improvement requirement is a food ingredient restriction requirement, the initial ingredients with restricted ingredients are screened out, and the remaining initial ingredients are added to the data group of the secondary screening ingredient set to form an improved formula set.

[0018] In the preferred embodiment of the knowledge graph-based food formulation generation method described above, the evaluation index set includes a nutritional suitability index, a comprehensive cost index, and a process feasibility index; wherein:

[0019] By analyzing the nutritional content data of each ingredient in the improved formula, the nutritional suitability index of each improved formula was obtained. The calculation formula is as follows:

[0020] ;

[0021] NAI stands for Nutritional Adaptation Index of the Improved Formula. This indicates the amount of the i-th ingredient used; This represents the content of the j-th nutrient in the i-th ingredient; i represents the sequence number of the ingredient in the improved formula, n represents the quantity of the ingredient; j represents the sequence number of the nutrient in the ingredient, and m represents the quantity of the nutrient. Indicates the daily reference for nutrient J; Let represent the weighting coefficient of nutrient j, and .

[0022] In the preferred embodiment of the knowledge graph-based food recipe generation method described above, the comprehensive cost index of each improved recipe is obtained by analyzing the cost data of each ingredient in the improved recipe. The calculation formula is as follows:

[0023] ;

[0024] CER represents the overall cost index of the improved formulation; This represents the net yield of the i-th ingredient. Let W represent the unit price of the i-th ingredient; W represent labor costs; and E represent the energy rate. This represents the time required to process the i-th type of ingredient.

[0025] In the preferred embodiment of the knowledge graph-based food formula generation method described above, the process feasibility index of each improved formula is obtained by analyzing the ingredient type and processing technology data of each ingredient in the improved formula. The calculation formula is as follows:

[0026] ,

[0027] PFE represents the feasibility index for improving the formulation and process; This represents the number of independent operation steps required for the i-th ingredient; This represents the ratio of the standard deviation of the processing parameter for the i-th ingredient to the target value; Represents a constant; Weighting coefficients representing the utilization rate of ingredients; Weighting coefficients representing processing complexity; Weighting coefficients representing process stability.

[0028] In the preferred embodiment of the knowledge graph-based food formulation generation method described above, a comprehensive formulation evaluation index is obtained by comprehensively analyzing the nutritional suitability index, the overall cost index, and the process feasibility index. The formula used is as follows:

[0029] ,

[0030] FCI stands for Formula Comprehensive Evaluation Index. The weighting coefficients represent the nutritional suitability index. The weighting coefficient represents the reciprocal of the comprehensive cost index; The weighting coefficient represents the process feasibility index.

[0031] In the preferred embodiment of the knowledge graph-based food formula generation method described above, the deviation index is calculated based on the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula. The formula used is as follows: Where PCZ represents the deviation index, A comprehensive evaluation index representing the improved formula; This represents the overall evaluation index of the initial formula;

[0032] Set a deviation index threshold, compare the deviation index with the deviation index threshold, and determine that the corresponding improved formula meets the standard when the deviation index is greater than or equal to the deviation index threshold.

[0033] (III) Beneficial Effects

[0034] This invention provides a food formula generation method based on knowledge graphs, which has the following beneficial effects:

[0035] (1) By constructing a food knowledge graph and two screening processes, it is possible to accurately select food that meets specific conditions from a large number of food ingredients according to the requirements of formula improvement. Compared with the traditional food selection method based on experience and subjective judgment, it greatly improves screening efficiency and accuracy and reduces research and development time and cost.

[0036] (2) By arranging and combining the ingredients in the improved formula set, a variety of improved formulas can be generated. This intelligent combination method can break through the limitations of human thinking, explore more possible combinations of ingredients, provide food R&D personnel with a broader space for innovation, and help develop more creative and market-competitive food products.

[0037] (3) Calculate the set of evaluation indices for improved formulas based on ingredient data, and comprehensively analyze them to obtain a comprehensive formula evaluation index, thus quantitatively evaluating the formulas from multiple dimensions. This makes the evaluation of food formulas more scientific, objective, and comprehensive. By comparing the comprehensive evaluation indices of the improved formulas with those of the initial formulas, a deviation index is obtained, which can accurately screen out improved formulas that meet the standards and are superior. Based on the analysis results of the deviation index, R&D personnel can clearly understand the differences between the improved formulas and the initial formulas, and make targeted adjustments and improvements to the formulas to continuously improve the quality and performance of food.

[0038] (4) The entire methodology is systematic and scientific, organically combining knowledge graphs, data processing, and evaluation analysis technologies to achieve automation and intelligence in food formula generation. From acquiring ingredient data and constructing knowledge graphs to screening and combining ingredients and evaluating and optimizing formulas, each step is closely linked and operates collaboratively, greatly improving the efficiency of food R&D, shortening the product development cycle, and enabling companies to respond to market demands more quickly and launch new food products. It can effectively support food companies in product innovation and differentiated competition. By rapidly generating and screening food formulas that meet specific requirements, companies can meet the diverse and personalized needs of consumers. Attached Figure Description

[0039] Figure 1 This is a schematic diagram illustrating the steps of a knowledge graph-based food formula generation method according to the present invention. Detailed Implementation

[0040] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0041] Example 1:

[0042] Please see Figure 1 This invention provides a food formula generation method based on knowledge graphs, comprising:

[0043] Step 1: Obtain ingredient data from known knowledge graphs, and construct an ingredient knowledge graph based on graph data group tools. Input the initial recipe into the ingredient knowledge graph, obtain substitute ingredients for different ingredients in the initial recipe, and form an initial set of selected ingredients.

[0044] The proposed method for generating food recipes is based on knowledge graphs. It involves collecting and preprocessing ingredient data to create an ingredient knowledge graph, which is then imported into a graph database for storage and management. Using knowledge embedding and representation learning tools, the structure and relationships of the ingredient knowledge graph are created, with each node representing an ingredient entity and its attributes, and edges representing relationships between entities. When an initial recipe is input, the method parses and identifies the ingredient names, queries the corresponding nodes in the knowledge graph and their associated quantitative data, and recommends alternatives for each ingredient based on similarity calculations, forming an initial set of selected ingredients.

[0045] Step 101: Obtaining ingredient data can include nutritional content, functional characteristics, flavor features, and cost data for different ingredients; the collected data will be cleaned, transformed, and preprocessed to form an ingredient knowledge graph.

[0046] It should be noted that ingredient data may also include processing technology data.

[0047] It should be noted that basic nutrient content data of food ingredients can be obtained through the USDA Food Composition Data Group, the Hong Kong Centre for Food Safety Data Group, professional nutritional component testing institutions, academic literature and research reports, and online data platforms and tools such as Open Food Facts. Functional characteristic data can be obtained by consulting professional journals in the fields of food science and chemical engineering, such as *Food Science*, *Food Chemistry*, and *Transactions of the Chinese Society of Agricultural Engineering*, to obtain research results and experimental data on functional characteristics of food ingredients, such as emulsifying properties, foaming properties, gelling properties, antioxidant properties, and water absorption properties. Flavor characteristic data can be obtained through flavor analysis instruments, such as gas chromatography-mass spectrometry (GC-MS), electronic nose, and electronic tongue, to perform qualitative and quantitative analysis of volatile flavor substances in food ingredients, and to determine the flavor components and their contents. Alternatively, reported flavor characteristic data and analysis results of food ingredients can be obtained by referring to flavor data groups and literature, such as professional flavor data groups like *FlavorChemicalsDatabase*, as well as academic literature related to food flavor research. Cost data can be obtained through industry reports and statistical data, such as food industry reports published by professional market research institutions like Euromonitor and Nielsen, to obtain analytical data on food market price trends and cost structures. In addition, one can refer to agricultural product price monitoring data published by government statistical departments or agricultural departments, or establish contact with food processing companies, catering companies, and other enterprises that directly purchase ingredients to obtain their ingredient procurement cost data, including bulk purchase prices and long-term contract prices.

[0048] Traditional food formulation methods suffer from limited and incomplete ingredient data sources, coupled with inconsistent data quality, often requiring extensive manual collection and processing, resulting in low efficiency. This new approach acquires ingredient data from multiple authoritative sources, including nutritional content, functional properties, flavor characteristics, and cost data, ensuring data richness and reliability. Simultaneously, the collected data undergoes preprocessing and transformation to create standardized ingredient knowledge graph data, improving data usability and quality and laying a solid foundation for subsequent formulation generation.

[0049] Step 102: Import the organized food ingredient knowledge graph data into a graph database such as Neo4j or OrientDB to store and manage the data. Utilize tools like OpenKE and TransX for knowledge embedding and representation learning, creating nodes and relationships. Each node represents an entity, such as food ingredient or nutritional components. Edges represent relationships between entities, such as "contains" or "has," forming the structure of the food ingredient knowledge graph. Within the knowledge graph, assign corresponding quantitative data as node attributes for each attribute of each food ingredient. For example, for an apple node, its nutritional components could include protein content (0.3g / 100g), fat content (0.2g / 100g), etc.; functional characteristics could include dietary fiber content (1.7g / 100g), etc.; flavor characteristics could include sweetness (approximately 10-15g sugar / 100g), etc.; and cost attributes could include price (3 yuan / 500g), etc.

[0050] In the past, the storage and management of ingredient data in food formulation development lacked systematicity and coherence, making it difficult to maximize the value of the data. By importing the organized ingredient knowledge graph data into graph databases such as Neo4j or OrientDB for storage and management, and utilizing tools like OpenKE and TransX for knowledge embedding and representation learning to create nodes and relationships, not only can efficient data storage be achieved, but also convenient querying, reasoning, and analysis of ingredient data can be facilitated. By setting quantified data for each attribute of each ingredient as node attributes, the characteristics of the ingredients are comprehensively and intuitively displayed, providing richer information support for food formulation development.

[0051] Step 103: When an initial recipe is input into the food ingredient knowledge graph, the initial recipe is first parsed to identify the name of each ingredient. Then, the nodes corresponding to these ingredient names are queried in the knowledge graph to obtain the quantitative data of each node and its associations. Based on the similarity calculation in the knowledge graph, by comparing the similarity of different ingredients in terms of nutritional components, functional characteristics, and flavor characteristics, several similar substitutes are recommended for each ingredient, and the quantitative data of the substitute ingredients are output to form an initial set of selected ingredients.

[0052] It should be noted that similarity calculation can be performed using models such as Euclidean distance calculation model, cosine similarity calculation model, or Jaccard similarity coefficient calculation model embedded in the food knowledge graph. This allows for the selection of ingredients with similarity thresholds as substitutes. Here, a known similarity calculation method can be chosen to calculate the similarity values ​​between the ingredients and those in the initial recipe in terms of nutritional content and functional characteristics. Appropriate weighting coefficients, such as 0.4 and 0.6, are then assigned to each similarity value to calculate the overall similarity value. This overall similarity value is then compared to a similarity threshold, which can be set to 80% or higher. When the overall similarity value is greater than or equal to the similarity threshold, the ingredient meets the selection criteria.

[0053] In current food formulation optimization, the selection of alternative ingredients often relies on experience, lacking scientific basis and comprehensive consideration. This makes it difficult to guarantee the similarity between alternative and original ingredients in multiple aspects, potentially leading to a decline in formulation performance. This proposed solution, when inputting an initial formulation into a food ingredient knowledge graph, identifies ingredient names through text parsing, queries corresponding nodes and their associated quantitative data within the knowledge graph, and calculates similarity based on the knowledge graph. It comprehensively considers the similarity of ingredients in nutritional components, functional properties, and flavor characteristics, recommending several similar substitutes for each ingredient to form an initial set of selected ingredients. This knowledge graph-based similarity calculation method can more accurately select suitable alternative ingredients, improving their quality and applicability, and providing a more reliable selection for food formulation optimization.

[0054] Step 104: Regularly collect new food ingredient data and research findings, and update the data in the knowledge graph to ensure the accuracy and timeliness of the information in the food composition database.

[0055] Research findings and market conditions in the food industry are constantly evolving, and ingredient data also needs to be updated promptly; otherwise, the accuracy and practicality of formula generation will be affected. This solution regularly collects new ingredient data and research results, updating the data in the knowledge graph to ensure the accuracy and timeliness of the food ingredient database. This allows food formula generation methods to adapt to market changes and new research findings, providing food companies with formula solutions that better meet their actual needs.

[0056] Step 2: Determine the recipe improvement requirements, and based on the requirements, select the ingredients that meet the requirements from the initial selection of ingredients to generate the secondary selection of ingredients. Then, merge the initial ingredients of the initial recipe with the secondary selection of ingredients to generate the improved recipe set. If one or more of the initial ingredients in the initial recipe do not have suitable replacement ingredients, then continue to use the initial ingredients of the initial recipe.

[0057] The proposed method for generating food recipes based on knowledge graphs involves a second screening of ingredients based on recipe improvement requirements after an initial screening. The initial ingredients are then combined with the screened ingredients to generate an improved recipe set. Specifically, the recipe improvement requirements are first determined, including cost constraints and ingredient composition constraints. Then, ingredients that meet the cost or composition constraints are selected from the initial screening set to generate a second-screened ingredient set. When generating the improved recipe set, if a suitable alternative to the initial ingredients exists, it is used; otherwise, the initial ingredients are retained.

[0058] Step 201: The recipe improvement requirements include cost restrictions and ingredient restriction requirements. When conducting secondary screening based on the recipe improvement requirements, if the restriction requirement is a cost restriction, the price cost of the initial ingredient is compared with the price cost of the substitute ingredient. If the price cost of the initial ingredient is greater than the price cost of the substitute ingredient, the substitute ingredient meets the requirement, and a secondary screening ingredient set is generated. If the recipe improvement requirement is an ingredient restriction requirement, which means that the ingredients of some ingredients are restricted, the substitute ingredients are not allowed to contain the restricted ingredients, and the ingredient data of the substitute ingredients that do not contain the restricted ingredients are used to generate a secondary screening ingredient set.

[0059] For example, food formulations for people with diabetes restrict the use of sugars such as white sugar, so ingredients containing white sugar are eliminated, and alternatives such as xylitol are selected; food formulations for people with lactose intolerance need to restrict the use of lactose, so ingredients such as raw milk are eliminated, and ingredients such as soy milk and lactose-free milk are selected.

[0060] Traditional food formulations face challenges in cost control, making it difficult to effectively reduce production costs while maintaining food quality. When improving formulations, companies often struggle to find lower-cost and suitable alternative ingredients, hindering cost optimization. Furthermore, when developing food formulations for specific populations (such as diabetics or lactose intolerant individuals), traditional methods often fail to accurately screen for ingredients free of restricted components, potentially leading to foods that do not meet the health needs of these groups or even pose health risks.

[0061] By setting cost limits, the price of initial ingredients is compared with the price of alternative ingredients. Alternative ingredients with lower costs than the initial ingredients are selected, generating a secondary selection set. This ensures that production costs are effectively reduced while meeting food quality requirements. For example, when multiple alternatives exist for a high-priced ingredient in the initial formula, cost comparison selects the cheaper alternative, thus optimizing cost control. By setting ingredient component limits, ingredients containing specific restricted components are eliminated, ensuring that alternative ingredients do not contain these restricted components. For example, when developing foods for diabetics, the use of sugars such as white sugar is restricted, and alternatives such as xylitol are selected; when developing foods for lactose-intolerant individuals, the use of lactose is restricted, and ingredients such as soy milk and lactose-free milk are selected. This meets the health needs of specific populations and improves the safety and suitability of food.

[0062] Step 202: The secondary selection of ingredients contains different data groups. Each data group corresponds to one ingredient in the initial recipe. The ingredients in this data group are all substitutes for the ingredients in the initial recipe. When one or more initial ingredients do not have substitutes, this initial ingredient is included in the data group of the secondary selection of ingredients.

[0063] Step 203: If the restriction requirement is a cost restriction requirement, add each of the initial ingredients in the initial formula to the data group of the secondary screening ingredient set to form an improved formula set; if the formula improvement requirement is a food ingredient restriction requirement, remove the initial ingredients with restricted ingredients, and add the remaining initial ingredients to the data group of the secondary screening ingredient set to form an improved formula set.

[0064] Traditional food formulation improvement methods often struggle to effectively integrate initial and alternative ingredients when generating improved formulations, especially when suitable substitutes are unavailable for some initial ingredients. This can compromise the feasibility and completeness of the improved formulation. A new approach addresses this by combining the initial ingredients of the initial formulation with a set of ingredients selected through secondary screening to generate a set of improved formulations. When one or more initial ingredients lack suitable substitutes, those same ingredients are retained, ensuring the completeness and feasibility of the improved formulation. For example, if a certain ingredient in the initial formulation has no suitable substitute after secondary screening, it is retained and combined with other substitutes to form the improved formulation, thus guaranteeing its successful application in actual production.

[0065] Step 3: Arrange and combine each ingredient in each data group in the improved recipe set to obtain several improved recipes.

[0066] By employing permutation and combination methods, ingredients in an improved recipe set can be comprehensively recombinated based on different ingredient pairing principles and culinary logic, resulting in several improved recipes. This not only greatly enriches the variety of food recipes but also satisfies the taste preferences and personalized needs of different consumers, providing food companies with more innovative options. This helps companies stand out in a highly competitive market and develop new products with unique flavors and nutritional value. This method effectively solves the problem of limited ingredient combinations and insufficient innovation caused by human experience limitations in traditional food recipe development. In the past, food researchers may have found it difficult to explore novel and unique ingredient combinations due to limitations in knowledge and experience. However, through permutation and combination methods, this limitation can be overcome, systematically generating a variety of possible recipe combinations and uncovering potential high-quality recipes.

[0067] Generating multiple improved formulas through permutations and combinations provides a richer sample base for subsequent formula evaluation and optimization. This allows researchers to more comprehensively analyze and compare different formulas, thereby more accurately selecting high-quality formulas that meet specific requirements and further improving the efficiency and quality of food formula development.

[0068] Step 4: Based on the ingredient data of various ingredients in different improved recipes, calculate the evaluation index set for each improved recipe, and conduct a comprehensive analysis based on the parameters in the evaluation index set to obtain the comprehensive evaluation index of the recipe.

[0069] It should be noted that the evaluation index set includes the nutritional suitability index, the overall cost index, and the process feasibility index.

[0070] Step 401: By analyzing the nutritional content data of each ingredient in the improved formula, the nutritional suitability index of each improved formula is obtained. The calculation formula is as follows:

[0071] ;

[0072] NAI stands for Nutritional Adaptation Index of the Improved Formula. This indicates the amount of the i-th ingredient used; This represents the content of the j-th nutrient in the i-th ingredient; i represents the index of the ingredient in the improved formula, which is a positive integer; n represents the quantity of the ingredient; j represents the index of the nutrient in the ingredient, which is a positive integer, such as protein, fat, carbohydrates, dietary fiber, etc.; m represents the quantity of the nutrient. The daily reference for nutrient J can be found in national or international nutritional standards; for example, 60g of protein / day, 60g of fat / day, etc. The weighting coefficient for nutrient j, reflecting its importance in the comprehensive index, can be adjusted according to actual needs. For example, when m takes values ​​from 1 to 4, it corresponds to four nutrients: protein, fat, carbohydrates, and dietary fiber, respectively. The possible values ​​for are: =0.3, =0.2, =0.3, =0.2.

[0073] It should be noted that when calculating the nutritional suitability index, the parameters involved in the formula need to be normalized to eliminate the dimensions of each parameter in order to facilitate the calculation.

[0074] Traditional food formulation nutritional assessments often focus on a single nutrient or a few nutrients, making it difficult to comprehensively and systematically evaluate the overall nutritional value of the food formulation. This limitation may result in insufficient nutritional balance in food formulations, failing to meet consumers' diverse needs for healthy foods.

[0075] By comprehensively considering the content and weighting coefficients of various nutrients, this approach can systematically evaluate the nutritional value of improved formulations. For example, when calculating the nutritional suitability index, it considers not only major nutrients such as protein, fat, and carbohydrates, but also other important nutrients such as dietary fiber. By assigning appropriate weighting coefficients to each nutrient, the importance of key nutrients can be highlighted, making the evaluation results more scientific and accurate. This contributes to the development of food formulations that better meet human nutritional needs.

[0076] The nutrient content varies greatly among different ingredients, and the units and reference standards for nutrients are also different, making direct comparison and analysis difficult. By using the standardized processing steps in the formula, the nutrient content in the ingredients is combined with the daily reference value, eliminating the differences in units and orders of magnitude between different nutrients. This allows the data of different nutrients to be compared and analyzed on the same scale, providing an intuitive understanding of the contribution of each nutrient in the formula and providing accurate data support for subsequent formula optimization.

[0077] Step 402: Analyze the cost data of each ingredient in the improved recipe to obtain the comprehensive cost index of each improved recipe. The calculation formula is as follows:

[0078] ;

[0079] CER represents the overall cost index of the improved formulation; This represents the net yield of the i-th ingredient, i.e., the proportion that can be used after processing. It can be determined by conducting multiple raw material processing experiments, calculating the net yield for each test, and taking the average of the results as the final net yield. The unit price of the i-th ingredient can be determined by referring to real-time market prices; W represents labor costs, which can be calculated based on payrolls or data from the human resources department; E represents the energy rate, which can be calculated from energy bills or equipment power consumption. This represents the time required to process the i-th ingredient, which can be obtained through kitchen operation time records or standardized process calculations.

[0080] It should be noted that when calculating the comprehensive cost index, the parameters involved in the formula need to be normalized to eliminate the dimensions of each parameter in order to facilitate the calculation.

[0081] Traditional food formulation cost assessments often focus solely on ingredient procurement costs, neglecting processing losses, labor, energy, and other cost factors, leading to incomplete and inaccurate cost evaluations. This new approach, by comprehensively considering factors such as ingredient yield, unit purchase price, labor costs, energy rates, and processing time, enables a comprehensive and accurate assessment of the actual costs of formula improvements. For example, when calculating the comprehensive cost index, it considers not only ingredient procurement costs but also processing losses and costs related to labor and energy, making the cost assessment more realistic and reliable. This helps food companies gain a more accurate understanding of product cost structure, providing strong support for pricing strategies and cost control.

[0082] Food companies often lack sophisticated management and analysis of cost data during cost management, making it difficult to identify key cost control points and potential cost-saving opportunities. By collecting and calculating the various cost data points in the formula, companies can achieve sophisticated management of cost data.

[0083] Step 403: Based on the analysis of the ingredient type and processing technology data of each ingredient in the improved formula, the process feasibility index of each improved formula is obtained. The calculation formula is as follows:

[0084] ,

[0085] PFE represents the feasibility index for improving the formulation and process; This represents the number of independent operation steps required for the i-th ingredient, such as washing, cutting, marinating, etc. This represents the ratio of the standard deviation of the processing parameter for the i-th ingredient to the target value. It can be one of the more important process parameters; for example, if the time scale of the process is important, time can be used as the processing parameter, and the standard deviation of the processing time can be calculated. If the processing temperature is more important, then temperature can be used as the processing parameter, and the standard deviation of the processing time can be calculated. The target value is the ideal value of the processing parameter. The smaller the value of , the more stable it is; To represent a constant, to prevent the denominator from being zero; Weighting coefficients representing the utilization rate of ingredients; Weighting coefficients representing processing complexity; The weighting coefficients represent the process stability; and The possible values ​​are: =0.3, =0.3, =0.4.

[0086] It should be noted that when calculating the process feasibility index, the parameters involved in the formula need to be normalized to eliminate the dimensions of each parameter in order to facilitate the calculation.

[0087] In food processing, the yield of different ingredients varies significantly, making it difficult to accurately assess and optimize ingredient utilization using traditional methods, which can easily lead to food waste and increased costs. By comparing the yield of an ingredient with its maximum yield and incorporating a weighting coefficient β1, the impact of each ingredient's utilization rate on the overall process feasibility can be quantitatively assessed. For example, if the yield of a certain ingredient is low, it indicates significant waste during processing. This can be addressed by optimizing the processing technology or finding alternative ingredients with higher utilization rates. This not only helps reduce food waste but also lowers production costs and improves the company's resource utilization efficiency.

[0088] The processing complexity varies significantly among different ingredients, making it difficult to comprehensively assess the impact of processing complexity on production efficiency and cost using traditional methods. This paper addresses this issue by comprehensively considering the number of independent operation steps and processing time required for each ingredient, combined with weighting coefficients. This allows for a quantitative assessment of the impact of the processing complexity of each ingredient on the overall feasibility of the process. For example, if an ingredient requires more independent processing steps and longer processing time, its processing complexity is high, which will reduce production efficiency and increase costs. Companies can improve production efficiency and reduce production costs by simplifying processing techniques, optimizing operational procedures, or selecting alternative ingredients with lower processing complexity.

[0089] In food processing, fluctuations in process parameters can affect product quality and production stability, and traditional methods are insufficient to effectively assess and control process stability. By analyzing the ratio of the standard deviation of processing parameters to the target value, and combining this with a weighting coefficient β3, the impact of the process stability of each ingredient on the overall process feasibility can be quantitatively assessed. For example, if the processing time or temperature of a certain ingredient fluctuates significantly, its process stability is poor, easily leading to unstable product quality. Enterprises can improve the stability and consistency of product quality by optimizing process parameter control, adopting more stable processing equipment, or selecting alternative ingredients with higher process stability.

[0090] In summary, the traditional food formulation development process lacks a comprehensive quantitative assessment of process feasibility, making it difficult to identify potential process problems during the formulation design stage, resulting in extended development cycles and increased production costs. The proposed solution, by comprehensively considering three key factors—ingredient utilization rate, processing complexity, and process stability—can comprehensively and systematically assess the process feasibility of improving the formulation.

[0091] Step 404: By comprehensively analyzing the nutritional compatibility index, overall cost index, and process feasibility index, a comprehensive formula evaluation index is obtained, based on the following formula:

[0092] ,

[0093] FCI stands for Formula Comprehensive Evaluation Index. The weighting coefficients represent the nutritional suitability index. The weighting coefficient represents the reciprocal of the comprehensive cost index; The weighting coefficients of the process feasibility index can be adjusted as needed. The possible values ​​are: =0.4, =0.3, =0.3.

[0094] It should be noted that when calculating the comprehensive evaluation index of the formula, the parameters involved in the formula need to be normalized to eliminate the dimensions of each parameter in order to facilitate the calculation of the formula.

[0095] Traditional food formulation evaluation methods often focus on a single dimension (such as nutritional value or cost), lacking a comprehensive assessment of multiple aspects such as nutrition, cost, and process feasibility. This results in incomplete evaluations that fail to meet the diverse needs of actual production. This new approach combines a nutritional suitability index, a comprehensive cost index, and a process feasibility index to comprehensively consider the performance of food formulations in three key areas: nutritional value, economic cost, and process feasibility. For example, when evaluating an improved formulation, it not only shows whether it meets nutritional requirements but also provides insights into its production costs and operational difficulties. This provides food companies with more comprehensive and valuable evaluation results, helping them make more informed decisions.

[0096] In the process of food formulation research and development and production decision-making, enterprises often face multiple risks, such as products failing to meet nutritional standards, exceeding production costs, or having infeasible processes. Traditional methods struggle to effectively identify and control these risks in the early stages. By using comprehensive evaluation indices, enterprises can identify potential risks in advance during the formulation design and optimization phases. For example, a formulation with a low FCI value may indicate significant problems in one of the following areas: nutrition, cost, or process feasibility. Enterprises can then make adjustments or optimizations in advance to avoid major problems in subsequent production. This not only helps reduce R&D and production risks but also reduces resource waste and improves the overall operational efficiency and market competitiveness of enterprises.

[0097] Step 5: Following the methods in Steps 1 to 4, calculate the comprehensive evaluation index of the initial formula. Analyze the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula to obtain the deviation index. Based on the deviation index, select the improved formulas that meet the standards, and sort the improved formulas that meet the standards according to the comprehensive evaluation index to generate a formula execution priority chart.

[0098] Step 501: Calculate the comprehensive evaluation index of the initial formula, referring to the methods in Steps 1 to 4.

[0099] Step 502: Calculate the deviation index based on the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula. The formula used is as follows: Where PCZ represents the deviation index, A comprehensive evaluation index representing the improved formula; This represents the overall evaluation index of the initial formula.

[0100] Step 503: Set the deviation index threshold, compare the deviation index with the deviation index threshold, and when the deviation index is greater than or equal to the deviation index threshold, determine that the corresponding improved formula meets the standard. Then, sort the improved formulas that meet the standard according to the comprehensive evaluation index in descending order of value to form the formula execution priority.

[0101] It should be noted that the deviation index threshold can be set to -10%. If you want to screen improved formulas with higher similarity, you can set the deviation index threshold to -5% to 5%. Or if you want to screen formulas with better overall performance, you can set the deviation index threshold to 10%.

[0102] Traditional methods lack scientific quantitative indicators and unified evaluation standards when screening improved formulations, making it difficult to quickly and accurately select the optimal solution from multiple improved formulations. This approach provides a quantitative method to measure the merits of an improved formulation relative to the initial formulation by calculating a deviation index. For example, if an improved formulation has a high deviation index, it indicates that its overall evaluation index is significantly better than the initial formulation, making it more worthy of priority consideration. This helps food companies quickly screen truly valuable solutions from numerous improved formulations, improving R&D efficiency and decision-making accuracy.

[0103] The evaluation results of different improved formulations are often difficult to compare directly, and the lack of a unified standardized evaluation framework leads to a high degree of subjectivity in the evaluation process. The deviation index calculation method standardizes the comparison between the evaluation results of the improved formulation and the initial formulation, eliminating the absolute value differences between different formulations, making the comparison between different improved formulations more intuitive and fair.

[0104] In actual production, companies need to choose from multiple improved formulations, but traditional methods struggle to comprehensively assess the risks and benefits of each formulation, easily leading to decision-making errors. By setting a deviation index threshold, companies can flexibly set screening criteria based on their risk tolerance and production goals. Only improved formulations with a deviation index reaching or exceeding the threshold are selected, effectively reducing the risk of choosing inefficient or inferior formulations. Simultaneously, ranking qualified improved formulations according to a comprehensive evaluation index provides companies with a clear decision-making basis, helping them prioritize the most likely successful formulations, thereby improving production success rates and market competitiveness.

[0105] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0106] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0107] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for generating food formulas based on knowledge graphs, characterized in that, include: Step 1: Obtain the ingredient data of the known knowledge graph, and construct the ingredient knowledge graph based on the graph data group tool. Input the initial recipe into the ingredient knowledge graph, obtain the substitute ingredients for different ingredients in the initial recipe, and form the initial set of ingredients for screening. Step 2: Determine the recipe improvement requirements, and based on the requirements, select the ingredients that meet the requirements from the initial ingredient set to generate the secondary ingredient set. Then, merge the initial ingredients of the initial recipe with the secondary ingredient set to generate the improved recipe set. Step 3: Arrange and combine each ingredient in each data group in the improved recipe set; Several improved formulas were obtained; Step 4: Based on the ingredient data of various ingredients in different improved recipes, calculate the evaluation index set for each improved recipe, and conduct a comprehensive analysis based on the parameters in the evaluation index set to obtain the comprehensive evaluation index of the recipe. Step 5: Following the methods in Steps 1 to 4, calculate the comprehensive evaluation index of the initial formula. Analyze the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula to obtain the deviation index. Based on the deviation index, select the improved formulas that meet the standards, and sort the improved formulas that meet the standards according to the comprehensive evaluation index to generate a formula execution priority chart. By comprehensively analyzing the nutritional compatibility index, overall cost index, and process feasibility index, a comprehensive formula evaluation index is obtained, based on the following formula: ； in, This represents the overall evaluation index of the formula. The weighting coefficients represent the nutritional suitability index. The weighting coefficient represents the reciprocal of the comprehensive cost index; The weighting coefficients represent the process feasibility index; This indicates the nutritional suitability index of the improved formula; In step five, the deviation index is calculated based on the comprehensive evaluation index of the improved formula and the comprehensive evaluation index of the initial formula. The formula used is as follows: ;in, Indicates the deviation index. A comprehensive evaluation index representing the improved formula; This represents the overall evaluation index of the initial formula; Set a deviation index threshold, compare the deviation index with the deviation index threshold, and determine that the corresponding improved formula meets the standard when the deviation index is greater than or equal to the deviation index threshold. Based on the cost data of each ingredient in the improved recipe, the comprehensive cost index of each improved recipe is obtained. The calculation formula is as follows: ； in, This represents the overall cost index of the improved formula; This represents the net yield of the i-th ingredient. Let W represent the unit price of the i-th ingredient; W represent labor costs; and E represent the energy rate. This represents the time required to process the i-th type of ingredient; Based on the analysis of the ingredient type and processing technology data of each ingredient in the improved formula, the process feasibility index of each improved formula is obtained. The calculation formula is as follows: ； in, Indicates the feasibility index for improving the formula and process; This represents the number of independent operation steps required for the i-th ingredient; This represents the ratio of the standard deviation of the processing parameter for the i-th ingredient to the target value; Represents a constant; Weighting coefficients representing the utilization rate of ingredients; Weighting coefficients representing processing complexity; Weighting coefficients representing process stability.

2. The food formula generation method based on knowledge graph according to claim 1, characterized in that, In step two, the recipe improvement requirements include cost restrictions and ingredient restriction requirements. When conducting secondary screening based on the recipe improvement requirements, if the restriction requirement is a cost restriction, the price cost of the initial ingredient is compared with the price cost of the substitute ingredient. If the price cost of the initial ingredient is greater than the price cost of the substitute ingredient, the substitute ingredient meets the requirement, and a secondary screening ingredient set is generated. If the recipe improvement requirement is an ingredient restriction requirement, which means that the composition of some ingredients is restricted, then the substitute ingredients are not allowed to contain the restricted ingredients, and the ingredient data of the substitute ingredients that do not contain the restricted ingredients are used to generate a secondary screening ingredient set.

3. The food formula generation method based on knowledge graph according to claim 2, characterized in that, In step two, the secondary screening ingredient set contains different data groups. Each data group corresponds to one ingredient in the initial recipe. The ingredients in this data group are all substitute ingredients for the ingredients in the initial recipe. When one or more initial ingredients do not have substitute ingredients, this initial ingredient is included in the data group of the secondary screening ingredient set.

4. The food formula generation method based on knowledge graph according to claim 3, characterized in that, In step two, if the constraint is a cost constraint, each initial ingredient in the initial formula is added to the data set of the secondary screening ingredient set to form an improved formula set. If the recipe improvement requirement is a restriction on the ingredients, the initial ingredients with restricted ingredients will be screened out, and the remaining initial ingredients will be added to the data set of the secondary screening ingredients to form the improved recipe set.

5. The food formula generation method based on knowledge graph according to claim 4, characterized in that, The evaluation index set includes the nutritional suitability index, the overall cost index, and the process feasibility index; among which: By analyzing the nutritional content data of each ingredient in the improved formula, the nutritional suitability index of each improved formula was obtained. The calculation formula is as follows: ； in, This indicates the nutritional suitability index of the improved formula; This indicates the amount of the i-th ingredient used; This represents the content of the j-th nutrient in the i-th ingredient; i represents the sequence number of the ingredient in the improved formula, n represents the quantity of the ingredient; j represents the sequence number of the nutrient in the ingredient, and m represents the quantity of the nutrient. This indicates the daily reference for nutrient J; This represents the weighting coefficient of nutrient j, and .

6. A food formula generation system based on knowledge graphs, characterized in that, This method is used to implement the knowledge graph-based food formulation generation method described in any one of claims 1-5.

Images