A method and system for constructing a nutrient knowledge graph and analyzing digestion of forage

Abstract

The present invention relates to the technical field of animal husbandry nutrition engineering, and in particular to a method and system for constructing a forage nutrition knowledge map and digestion analysis. The method comprises the following steps: constructing a forage nutrition knowledge map based on forage literature resource data; identifying key nutrients according to the forage nutrition knowledge map, and generating forage key nutrient data; identifying anti-nutritional action relationship paths according to the forage key nutrient data, and generating anti-nutritional interaction path data; extracting digestion influencing factors according to the anti-nutritional interaction path data, and constructing a nutrient digestion prediction model to generate a nutrient digestion prediction model; predicting nutrient digestibility using the nutrient digestion prediction model to generate a nutrient predicted digestibility; optimizing feeding formulas according to the nutrient predicted digestibility to obtain feeding effect evaluation data. The present invention realizes precision feeding plan customization through the construction of forage nutrition knowledge maps.

Description

Technical Field

[0001] The present invention relates to the technical field of animal husbandry nutrition engineering, and in particular to a method and system for constructing a forage nutrition knowledge map and performing digestion analysis. Background Art

[0002] With the rapid development of animal husbandry, the nutritional value and digestion and metabolism of forage have received more and more attention. The nutritional components of forage directly affect the growth and development, health status and product quality of livestock. At the same time, different types of livestock have different nutritional requirements for forage. Therefore, it is of great significance to fully and accurately master the nutritional knowledge of forage and make reasonable proportions and feeding management according to the actual needs of livestock to improve the efficiency of animal husbandry. Traditional forage digestion analysis methods often only focus on a single nutritional indicator, ignore the complex relationship and interaction between nutrients, and mainly rely on laboratory analysis and animal experiments. These methods are often time-consuming and costly, and it is difficult to fully reflect the nutritional characteristics and digestion laws of forage. Summary of the invention

[0003] Based on this, the present invention provides a method and system for constructing a forage nutrition knowledge map and digestion analysis to solve at least one of the above technical problems.

[0004] To achieve the above purpose, a method for constructing a forage nutrition knowledge graph and digestion analysis includes the following steps:

[0005] Step S1: Acquire forage literature resource data; construct a forage nutrition knowledge map based on the forage literature resource data;

[0006] Step S2: screening nutrient component nodes according to the forage nutrition knowledge map to obtain forage nutrient component node data; performing co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; identifying key nutrient components according to the nutrient component co-occurrence matrix to generate forage key nutrient component data;

[0007] Step S3: extracting anti-nutritional factors according to the key nutrient data of forage to generate anti-nutritional factor data; analyzing the action relationship pattern of the key nutrient data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identifying the anti-nutritional action relationship path according to the interaction relationship pattern data to generate anti-nutritional interaction path data;

[0008] Step S4: extracting digestion influencing factors according to the anti-nutritional interaction pathway data to generate digestion influencing factor data; constructing a nutrient digestion prediction model according to the digestion influencing factor data to generate a nutrient digestion prediction model;

[0009] Step S5: using the nutrient digestion prediction model to predict the digestibility of the nutrient components of the acquired target forage sample, and generating the predicted digestibility of the nutrient components; performing a comprehensive nutritional value score based on the predicted digestibility of the nutrient components, and generating the nutritional value data of the forage;

[0010] Step S6: Optimizing the feeding formula according to the nutrient value data of the forage to generate optimized feeding plan data; evaluating the feeding effect according to the optimized feeding plan data to obtain feeding effect evaluation data.

[0011] The present invention integrates scattered forage nutritional information into a structured knowledge network, which is helpful for rapid retrieval and association analysis. The construction of the knowledge graph makes the complex relationship between the nutritional components of forage clear and visible, and the complex relationship between entities such as forage, nutrients, and anti-nutritional factors can be intuitively displayed through the knowledge graph. By performing node screening and co-occurrence frequency analysis on the forage nutritional knowledge graph, the data complexity is effectively reduced and the most important nutritional elements are highlighted. The generation of the co-occurrence matrix reveals the correlation between different nutrients, and the potential correlation between different nutrients can be explored, providing a more comprehensive perspective for the evaluation of the nutritional value of forage. Taking the influence of anti-nutritional factors into consideration, the nutritional value of forage can be evaluated more comprehensively and objectively, and the complex relationship of beneficial nutrients in forage is deeply explored. By identifying the anti-nutritional action pathway, this information has important guiding significance for preventing and reducing the negative impact of anti-nutritional factors in forage. By constructing a nutritional digestion prediction model, the digestibility of different forages in different animals can be predicted more accurately, overcoming the shortcomings of traditional methods that rely on empirical data and have low prediction accuracy. Through prediction, the expected digestibility of each nutrient component of forage can be obtained before actual feeding, which is crucial for optimizing feed formula and improving feeding efficiency. The prediction results can help breeders avoid using feed with low digestibility, thereby reducing feed waste and improving the nutrient absorption efficiency of animals. Through the prediction of nutrient digestibility and comprehensive scoring of nutritional value, the nutritional value of the target forage can be evaluated more objectively and comprehensively. The nutritional value of the same forage can be personalized according to the nutritional needs of different animals, and accurate feeding plans can be formulated for different animals. The optimized formula can maximize the nutritional value of various forages, while taking into account cost-effectiveness, and provide animals with the best nutritional support. This optimization not only improves feed utilization efficiency, but also meets the nutritional needs of animals, promotes the healthy growth of animals, and improves the production performance and product quality of animals. Therefore, a forage nutrition knowledge map construction and digestion analysis method of the present invention is obtained through forage literature resources to construct a complete forage nutrition knowledge map. By identifying key nutrients and analyzing anti-nutritional factors, the complex relationships and interactions between nutrients are taken into account to quickly evaluate the nutritional value of different forages. By constructing a nutrient digestion prediction model, the digestion patterns of forages in animals are accurately predicted, and the feeding plan is customized according to the needs of animals at different growth stages and physiological states, thus achieving precise and personalized customization of feeding plans.

[0012] Preferably, step S1 comprises the following steps:

[0013] Step S11: Acquire forage literature resource data;

[0014] Step S12: performing data semantic cleaning on the forage literature resource data and performing text data deduplication processing to generate original forage corpus data;

[0015] Step S13: semantically annotating the original forage corpus data and performing named entity recognition to generate forage entity corpus data, wherein the forage entity data includes the forage name, nutritional components, anti-nutritional factors, and feeding animal species;

[0016] Step S14: performing ontology matching calculation on the forage entity corpus data and performing knowledge entity merging processing to generate forage entity data;

[0017] Step S15: extracting entity association relationships from the original forage corpus data through the forage entity data to generate entity association relationship data;

[0018] Step S16: constructing a knowledge graph based on the entity association relationship data and the forage entity data to generate an initial forage nutrition knowledge graph;

[0019] Step S17: performing forage knowledge link expansion processing on the forage literature resource data through the initial forage nutrition knowledge graph to obtain a forage nutrition knowledge graph.

[0020] The present invention can cover the latest research results and practical experience in the field of forage by collecting a large number of relevant documents. Through data semantic cleaning, noise data is removed, data quality is significantly improved, irrelevant information and repeated content are removed, and data consistency and comprehensibility are ensured. Unstructured text is converted into structured information, key entities such as forage name, nutritional components, anti-nutritional factors and feeding animal varieties are clearly identified, and the accuracy and efficiency of subsequent analysis are improved. Through ontology matching and entity merging, entity ambiguity caused by problems such as irregular naming and synonyms can be eliminated, and the uniqueness and accuracy of entities in the knowledge graph can be ensured. Entity merging processing reduces redundancy and improves data consistency and availability. Through association relationship extraction, the relationship between different components of forage, the relationship between forage and animals, etc. can be revealed. The nutritional knowledge of forage is presented in the form of a map, so that people can more intuitively understand the complex relationship between forage, nutritional components, anti-nutritional factors and animals. Through knowledge link expansion, new entities and relationships can be continuously extracted from new literature data and added to the knowledge map, making the content of the knowledge map richer and more complete.

[0021] Preferably, step S2 comprises the following steps:

[0022] Step S21: screening nutrient component nodes according to the forage nutrient knowledge graph to obtain forage nutrient component node data;

[0023] Step S22: performing co-occurrence frequency statistical processing according to the nutrient component node data of forage to generate nutrient co-occurrence frequency data;

[0024] Step S23: constructing a nutrient component co-occurrence matrix for the nutrient component node data of the forage by using the nutrient co-occurrence frequency data to generate a nutrient component co-occurrence matrix;

[0025] Step S24: performing singular value decomposition processing on the nutrient co-occurrence matrix and mining potential association relationships to generate nutrient potential association data;

[0026] Step S25: calculating the association strength of the potential association data of the nutrient components through the nutrient co-occurrence frequency data based on a preset support vector machine to obtain potential association strength data;

[0027] Step S26: inferring the type of association relationship based on the potential association strength data to generate nutrient component association type data;

[0028] Step S27: performing importance node evaluation on the forage nutrient component node data through the nutrient component association type data, and identifying key nutrient components to generate forage key nutrient component data.

[0029] The present invention screens the nutrient nodes of the forage nutrition knowledge map, effectively reduces the data complexity, and focuses on the most relevant nutrient information. The co-occurrence frequency statistical processing of the nutrient node data of the forage reveals the degree of association between different nutrients, and provides a quantitative basis for identifying important nutrient combinations. The co-occurrence frequency data reflects the distribution characteristics of the nutrients in the forage, which helps to understand the interdependence of different nutrients. The relationship between the nutrients is converted into a visual and computable matrix form, and the co-occurrence matrix provides a global view of the relationship between the nutrients. The singular value decomposition processing and potential association mining of the nutrient co-occurrence matrix can reveal the main patterns and structures in the data, while the potential association mining can discover the nutrient interactions that are not obvious on the surface. The support vector machine can effectively process high-dimensional data and nonlinear relationships, and can more accurately quantify the association strength between different nutrients. The numerical association strength is converted into a specific relationship type, such as synergy, antagonism, etc. The identification of the association type helps to deeply understand the mutual influence mechanism between nutrients. The identification of key nutrients can help people grasp the key points and focus on the factors that have the greatest impact on the nutritional value of forage.

[0030] Preferably, step S3 comprises the following steps:

[0031] Step S31: extracting anti-nutritional factors based on the key nutritional component data of forage through the forage nutrition knowledge map to generate anti-nutritional factor data;

[0032] Step S32: performing principal component coupling strength analysis on the key nutrient component data of forage through the anti-nutritional factor data to generate anti-nutritional factor-nutrient component coupling pair data;

[0033] Step S33: performing spectral clustering processing on the anti-nutrient factor-nutrient component coupling data, and performing action relationship pattern analysis to generate interaction relationship pattern data, wherein the interaction relationship pattern data includes synergistic action pattern, antagonistic action pattern and cumulative action pattern;

[0034] Step S34: conducting literature evidence mining on the interaction relationship pattern data, and quantifying the strength of the interaction relationship to generate interaction relationship strength data;

[0035] Step S35: Refining the interaction relationship type of the anti-nutritional factor-nutrient coupling data through the interaction relationship strength data, and constructing the interaction network to generate the anti-nutritional factor-nutrient interaction network;

[0036] Step S36: Identify anti-nutritional action relationship pathways based on the anti-nutritional factor-nutrient interaction network to generate anti-nutritional interaction pathway data.

[0037] The present invention performs query and association analysis in the constructed forage nutrition knowledge map, and extracts the anti-nutritional factor information that has potential interactions with these key nutrients, such as tannins, phytic acid, saponins, etc. The principal component coupling strength analysis of the key nutrient data of forage is performed through the anti-nutritional factor data, which reflects the interaction between anti-nutritional factors and key nutrients, and reveals which anti-nutritional factors have a significant effect on the absorption of specific nutrients. The complex coupling relationship is classified into different modes such as synergy, antagonism and cumulative effect, which can more clearly understand different types of nutrition-anti-nutritional interaction mechanisms. Combining theoretical analysis with actual research evidence improves the reliability and practicality of the analysis results. Through literature evidence mining, the results of pattern analysis can be verified and supplemented. Through the interaction relationship strength data, the action relationship between anti-nutritional factors and nutrients can be more finely portrayed, such as distinguishing different degrees of antagonism, which not only shows the relationship between a single anti-nutritional factor and nutrients, but also reveals the complex interaction between multiple factors. By identifying the anti-nutritional action relationship pathway, we can gain a deeper understanding of the specific mechanism by which anti-nutritional factors affect the nutritional value of forage, such as how an anti-nutritional factor inhibits the activity of a certain enzyme and ultimately reduces the digestibility of a certain nutrient.

[0038] Preferably, step S4 comprises the following steps:

[0039] Step S41: mining nutritional digestion knowledge based on the key nutritional component data of forage to generate nutritional digestion knowledge data;

[0040] Step S42: performing nutrient digestion characteristic analysis on the key nutrient component data of the forage using the nutrient digestion knowledge data to obtain nutrient digestion characteristic data;

[0041] Step S43: extracting digestion influencing factors from the nutrient digestion characteristic data based on the anti-nutritional interaction pathway data to generate digestion influencing factor data;

[0042] Step S44: performing digestion characteristic integration on the digestion influencing factor data and the nutrient digestion characteristic data to generate nutrient digestion characteristic data;

[0043] Step S45: constructing a nutrient digestion prediction model based on the nutrient digestion characteristic data to generate a nutrient digestion prediction model.

[0044] The present invention can obtain key information such as the digestion rules, absorption mechanism and utilization efficiency of various nutrients in animals through knowledge mining. Combining general digestion knowledge with specific forage nutrients reveals the unique digestion characteristics of each nutrient in a specific forage. This analysis helps to understand the behavior patterns of different nutrients in the digestion process, such as digestion rate, absorption site, metabolic pathway, etc. The influence of anti-nutritional factors is incorporated into the digestion characteristic analysis, and various factors affecting nutrient digestion are fully considered. By extracting these digestion influencing factors, researchers can more accurately predict and explain the nutrient digestion effects under different conditions. Integrating the digestion influencing factor data and nutrient digestion characteristic data into digestion characteristics can obtain a more complete and accurate description of the digestion process, including positive nutrient absorption characteristics and potential negative impact factors. A nutrient digestion prediction model is constructed based on the nutrient digestion characteristic data, and the constructed prediction model can simulate complex digestion processes and predict the digestion effects of different forages in different animals.

[0045] Preferably, step S45 includes the following steps:

[0046] Step S451: Obtaining animal digestibility experimental data of forage;

[0047] Step S452: performing feature engineering processing on the experimental data of digestibility of forage animals to generate feature data of digestibility of forage animals;

[0048] Step S453: mapping the nutritional components and the digestibility of the forage using a random forest regression algorithm based on the forage animal digestibility characteristic data, and constructing a preliminary nutrient digestion prediction model;

[0049] Step S454: randomly dividing the nutrient digestion characteristic data into data sets to obtain a nutrient digestion training set and a nutrient digestion verification set;

[0050] Step S455: training the preliminary nutrition digestion prediction model using the nutrition digestion training set, and performing hyperparameter tuning according to the Michaelis equation to obtain an optimized nutrition digestion prediction model;

[0051] Step S456: Performing model verification processing on the optimized nutritional digestion prediction model through the nutritional digestion verification set to obtain the nutritional digestion prediction model.

[0052] The experimental data of digestibility of forage animals obtained by the present invention reflects the digestion of forage in animals under real conditions, and includes the comprehensive effect of various influencing factors. The experimental data of digestibility of forage animals are subjected to feature engineering processing, and the original experimental data are converted into a feature set with more representativeness and predictive value, reflecting the key factors of digestibility changes, and reducing the noise and redundancy of the data. The complex nonlinear relationship is converted into a computable model. The use of the random forest algorithm can effectively process high-dimensional features, capture complex feature interactions, and has strong robustness to outliers. Through random partitioning, data distribution deviation can be avoided, so that the model can learn more general laws. Iterative learning continuously improves the performance of the model, which helps to find the best model configuration. This optimization process not only improves the prediction accuracy of the model, but also enhances the adaptability of the model to different forages and animal species. By using an independent validation set, the performance of the model on unseen data can be evaluated, thereby judging the generalization ability of the model. This validation process can not only confirm the prediction accuracy of the model, but also discover potential overfitting or underfitting problems.

[0053] Preferably, step S5 comprises the following steps:

[0054] Step S51: obtaining a target forage sample; performing feed sample informatization processing on the target forage sample to obtain digital forage sample data;

[0055] Step S52: using the forage nutrition knowledge map to match the digitized forage sample data with standard forage data to obtain matching forage nutrition data;

[0056] Step S53: performing sample-specific nutritional analysis according to the matched forage nutritional data to generate forage-specific nutritional data;

[0057] Step S54: performing environmental impact index analysis based on the forage-specific nutritional data to obtain a forage environmental adjustment coefficient;

[0058] Step S55: using a nutrient digestibility prediction model to predict nutrient digestibility of the digital forage sample data based on the forage environment adjustment coefficient, and generating a predicted nutrient digestibility;

[0059] Step S56: Calculate the digestible amount of nutrients according to the predicted digestibility of the nutrients to obtain the digestible amount data of the nutrients;

[0060] Step S57: Comprehensively score the nutritional value of the digitized forage sample data using the nutrient digestible amount data to generate forage nutritional value data.

[0061] The present invention converts actual forage samples into a digital form that can be analyzed by a computer. Informatization processing not only improves the standardization of data, but also greatly increases the efficiency and accuracy of data processing. The specific sample data is associated with a systematic knowledge base, which greatly enriches the nutritional information of the sample. Through this matching, the nutritional component information that is not directly measured in the sample can be quickly obtained to fill the data gap. Through specific analysis, the nutritional components in the sample that are different from the standard or average level can be identified, which is crucial for evaluating the advantages and potential risks of specific forage. The impact of environmental factors on the nutritional value of forage is taken into account, so that the evaluation results are closer to the actual application scenario. The introduction of the environmental adjustment coefficient enables the nutritional prediction model to adapt to different geographical, climatic and feeding conditions, and improves the accuracy and applicability of the prediction. Using the prediction model and the environmental adjustment coefficient, the nutrient digestibility of the target forage sample under specific environmental conditions can be more accurately predicted. The predicted digestibility is converted into the amount of nutrients that can actually be absorbed and utilized by animals, and the digestible amount data intuitively reflects the actual utilization value of each nutrient in the forage. The comprehensive nutritional value score can convert complex nutritional information into simple and intuitive indicators, making it convenient for farmers and feed companies to quickly evaluate and compare the nutritional value of different forages.

[0062] Preferably, step S6 comprises the following steps:

[0063] Step S61: accurately dividing the growth stages of the target animals to obtain the growth stage data of the animals;

[0064] Step S62: quantifying the production performance target of the animal according to the growth stage data of the raised animal to obtain the production performance target data;

[0065] Step S63: Calculating the nutritional component requirement according to the production performance target data to generate animal nutritional component requirement data;

[0066] Step S64: optimizing the feeding formula of the target animal based on the nutrient value data of the forage and using the animal nutrient requirement data to generate optimized feeding plan data;

[0067] Step S65: formulating an accurate nutrition supply plan according to the optimized feeding plan data, and collecting feeding feedback data to obtain feeding feedback data;

[0068] Step S66: Perform feeding effect evaluation according to the feeding feedback data to obtain feeding effect evaluation data.

[0069] The present invention can more accurately grasp the specific needs of animals at different stages of development by accurately dividing the growth stages, so as to formulate more targeted feeding strategies. Abstract growth goals are converted into specific and measurable indicators. By quantifying production performance goals, breeders can set clear interim goals, such as daily weight gain, milk production, egg production rate, etc. The production goals are converted into specific nutritional needs, laying the foundation for precise feeding. By calculating the specific nutritional needs under different growth stages and production goals, it can be ensured that the feeding plan can accurately meet the physiological needs of animals. Through the optimization of feeding formulas, efficient utilization of feed resources can be achieved, feeding costs can be reduced, while meeting the nutritional needs of animals and improving production performance. The formulation of precise nutrition supply plans ensures that the optimized formulas can be accurately applied to daily feeding. Through feeding effect evaluation, problems in the feeding plan can be discovered in a timely manner, and corresponding adjustments and optimizations can be made to continuously improve the level of feeding management and achieve the best production benefits.

[0070] Preferably, step S64 includes the following steps:

[0071] Step S641: extracting the real-time forage supply of the target animal to obtain the real-time forage supply data;

[0072] Step S642: performing a nutrient demand gap analysis on the real-time forage supply data using the animal nutrient component demand data to obtain nutrient demand gap data;

[0073] Step S643: using the forage nutritional value data to match the nutritional demand gap data to compensate for the forage nutritional gap, and generating forage nutritional compensation data;

[0074] Step S644: Processing the availability of raw materials according to the nutrient compensation data of forage, and generating available feed raw material list data;

[0075] Step S645: generating a forage ratio scheme for the real-time forage supply data based on the available feed raw material list data to obtain forage ratio scheme data;

[0076] Step S646: Perform precise nutrient supply balance processing on the forage and grass ratio program data, and perform multi-objective optimization processing to generate optimized feeding program data.

[0077] The present invention obtains the current forage supply of the target animal in real time, and the real-time forage supply data reflects the actual feed intake of the animal, which helps to timely discover and solve problems in the feeding process, such as abnormal feed intake, feed waste, etc. The theoretical demand is compared with the actual intake, and the deficiencies in nutrient supply are accurately identified. The nutrient demand gap data intuitively reflects the aspects that need to be improved in the current feeding plan, which helps to prevent the problem of malnutrition or excess, ensure that the animal obtains a balanced nutrient supply, thereby optimizing growth performance and improving the health level of the animal. By accurately matching the nutritional value of forage with the demand gap of the animal, the most suitable nutritional supplement strategy can be formulated. This precise matching not only improves the feed utilization efficiency, but also minimizes nutrient waste and reduces feeding costs. The ideal nutritional supplement scheme is combined with the actual available resources to ensure the executability of the feeding scheme. By optimizing the ratio, the synergistic effect of multiple feed raw materials can be achieved, and the overall nutritional value and digestibility can be improved. The multi-objective optimization process takes into account multiple aspects such as cost-effectiveness, environmental impact, and animal welfare, so that the final feeding plan is not only optimal in nutrition, but also can achieve a balance in terms of economy and sustainability.

[0078] The present invention also provides a forage nutrition knowledge graph construction and digestion analysis system, which executes the forage nutrition knowledge graph construction and digestion analysis method as described above, and the forage nutrition knowledge graph construction and digestion analysis system comprises:

[0079] A knowledge graph construction module is used to obtain forage literature resource data; and to construct a forage nutrition knowledge graph based on the forage literature resource data;

[0080] The nutrient component identification module is used to screen nutrient component nodes according to the forage nutrient knowledge map to obtain forage nutrient component node data; perform co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; perform key nutrient component identification according to the nutrient component co-occurrence matrix to generate forage key nutrient component data;

[0081] The anti-nutritional interaction analysis module is used to extract anti-nutritional factors based on the key nutritional component data of forage and generate anti-nutritional factor data; analyze the action relationship pattern of the key nutritional component data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identify the anti-nutritional action relationship path based on the interaction relationship pattern data to generate anti-nutritional interaction path data;

[0082] The digestion model construction module is used to extract digestion influencing factors based on the anti-nutritional interaction pathway data and generate digestion influencing factor data; construct a nutrient digestion prediction model based on the digestion influencing factor data and generate a nutrient digestion prediction model;

[0083] The nutritional value assessment module is used to predict the digestibility of the nutrient components of the acquired target forage samples using the nutritional digestion prediction model to generate the predicted digestibility of the nutrient components; perform a comprehensive nutritional value score based on the predicted digestibility of the nutrient components to generate the nutritional value data of the forage;

[0084] The intelligent feeding plan optimization module is used to optimize the feeding formula according to the nutritional value data of forage and generate optimized feeding plan data; and to evaluate the feeding effect according to the optimized feeding plan data to obtain the feeding effect evaluation data. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 It is a schematic diagram of the steps of the forage nutrition knowledge map construction and digestion analysis method and system of the present invention;

[0086] Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart;

[0087] Figure 3 for Figure 1 Detailed implementation steps of step 53 are shown in the flowchart;

[0088] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0089] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0090] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0091] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0092] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for constructing a forage nutrition knowledge map and digestion analysis, comprising the following steps:

[0093] Step S1: Acquire forage literature resource data; construct a forage nutrition knowledge map based on the forage literature resource data;

[0094] Step S2: screening nutrient component nodes according to the forage nutrition knowledge map to obtain forage nutrient component node data; performing co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; identifying key nutrient components according to the nutrient component co-occurrence matrix to generate forage key nutrient component data;

[0095] Step S3: extracting anti-nutritional factors according to the key nutrient data of forage to generate anti-nutritional factor data; analyzing the action relationship pattern of the key nutrient data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identifying the anti-nutritional action relationship path according to the interaction relationship pattern data to generate anti-nutritional interaction path data;

[0096] Step S4: extracting digestion influencing factors according to the anti-nutritional interaction pathway data to generate digestion influencing factor data; constructing a nutrient digestion prediction model according to the digestion influencing factor data to generate a nutrient digestion prediction model;

[0097] Step S5: using the nutrient digestion prediction model to predict the digestibility of the nutrient components of the acquired target forage sample, and generating the predicted digestibility of the nutrient components; performing a comprehensive nutritional value score based on the predicted digestibility of the nutrient components, and generating the nutritional value data of the forage;

[0098] Step S6: Optimizing the feeding formula according to the nutrient value data of the forage to generate optimized feeding plan data; evaluating the feeding effect according to the optimized feeding plan data to obtain feeding effect evaluation data.

[0099] The present invention integrates scattered forage nutritional information into a structured knowledge network, which is helpful for rapid retrieval and association analysis. The construction of the knowledge graph makes the complex relationship between the nutritional components of forage clear and visible, and the complex relationship between entities such as forage, nutrients, and anti-nutritional factors can be intuitively displayed through the knowledge graph. By performing node screening and co-occurrence frequency analysis on the forage nutritional knowledge graph, the data complexity is effectively reduced and the most important nutritional elements are highlighted. The generation of the co-occurrence matrix reveals the correlation between different nutrients, and the potential correlation between different nutrients can be explored, providing a more comprehensive perspective for the evaluation of the nutritional value of forage. Taking the influence of anti-nutritional factors into consideration, the nutritional value of forage can be evaluated more comprehensively and objectively, and the complex relationship of beneficial nutrients in forage is deeply explored. By identifying the anti-nutritional action pathway, this information has important guiding significance for preventing and reducing the negative impact of anti-nutritional factors in forage. By constructing a nutritional digestion prediction model, the digestibility of different forages in different animals can be predicted more accurately, overcoming the shortcomings of traditional methods that rely on empirical data and have low prediction accuracy. Through prediction, the expected digestibility of each nutrient component of forage can be obtained before actual feeding, which is crucial for optimizing feed formula and improving feeding efficiency. The prediction results can help breeders avoid using feed with low digestibility, thereby reducing feed waste and improving the nutrient absorption efficiency of animals. Through the prediction of nutrient digestibility and comprehensive scoring of nutritional value, the nutritional value of the target forage can be evaluated more objectively and comprehensively. The nutritional value of the same forage can be personalized according to the nutritional needs of different animals, and accurate feeding plans can be formulated for different animals. The optimized formula can maximize the nutritional value of various forages, while taking into account cost-effectiveness, and provide animals with the best nutritional support. This optimization not only improves feed utilization efficiency, but also meets the nutritional needs of animals, promotes the healthy growth of animals, and improves the production performance and product quality of animals. Therefore, a forage nutrition knowledge map construction and digestion analysis method of the present invention is obtained through forage literature resources to construct a complete forage nutrition knowledge map. By identifying key nutrients and analyzing anti-nutritional factors, the complex relationships and interactions between nutrients are taken into account to quickly evaluate the nutritional value of different forages. By constructing a nutrient digestion prediction model, the digestion patterns of forages in animals are accurately predicted, and the feeding plan is customized according to the needs of animals at different growth stages and physiological states, thus achieving precise and personalized customization of feeding plans.

[0100] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a forage nutrition knowledge graph construction and digestion analysis method of the present invention. In this embodiment, the forage nutrition knowledge graph construction and digestion analysis method includes the following steps:

[0101] Step S1: Acquire forage literature resource data; construct a forage nutrition knowledge map based on the forage literature resource data;

[0102] In an embodiment of the present invention, a literature database such as PubMed, Web of Science or CNKI is used to search with keywords such as "forage", "nutrition", "ingredients", "digestibility", etc. to collect relevant literature resources. Text mining and information extraction are performed on the collected literature to identify key information such as the type, nutritional components, and growth environment of the forage. Then, the extracted information is structured using natural language processing technology to form triples of entities and relationships. For example, by performing keyword extraction on the collected literature, it is found that the words "grass", "soybean meal", "corn", "crude protein", "crude fiber", "digestibility" and the like appear frequently and are correlated with each other. Then, the triple data is imported and the association between entities is established, and a graph database such as Neo4j is used to construct a knowledge graph of forage nutrition.

[0103] Step S2: screening nutrient component nodes according to the forage nutrition knowledge map to obtain forage nutrient component node data; performing co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; identifying key nutrient components according to the nutrient component co-occurrence matrix to generate forage key nutrient component data;

[0104] In an embodiment of the present invention, nodes related to "nutrients" are screened out based on the forage nutrition knowledge graph, such as "crude protein", "crude fat", "crude fiber", "ash", "water", etc. The co-occurrence frequency statistics are performed on the screened nutrient node data, the frequency of simultaneous appearance of each nutrient in the literature is calculated, and a nutrient co-occurrence matrix is ​​generated. The co-occurrence frequency statistics are performed on the screened nutrient node data, the frequency of simultaneous appearance of each nutrient in the literature is calculated, and a nutrient co-occurrence matrix is ​​generated. Based on the nutrient co-occurrence matrix, key nutrients can be identified. For example, if a nutrient has a high co-occurrence frequency with multiple other nutrients in the co-occurrence matrix, it means that the nutrient plays an important role in the nutritional value of forage, and it can be identified as a key nutrient.

[0105] Step S3: extracting anti-nutritional factors according to the key nutrient data of forage to generate anti-nutritional factor data; analyzing the action relationship pattern of the key nutrient data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identifying the anti-nutritional action relationship path according to the interaction relationship pattern data to generate anti-nutritional interaction path data;

[0106] In an embodiment of the present invention, literature retrieval or database query is used to extract anti-nutritional factors related to key nutrient data and generate anti-nutritional factor data. For example, if "crude protein" is identified as a key nutrient, anti-nutritional factors related to it, such as "trypsin inhibitors" and "lectins", etc., can be extracted. The key nutrient data and the anti-nutritional factor data are subjected to association analysis to identify the influence pattern of anti-nutritional factors on key nutrients and generate interaction relationship pattern data. For example, the influence of "trypsin inhibitors" on the digestibility of "crude protein" can be analyzed and its influence pattern can be identified. Based on the interaction relationship pattern data, the action relationship path of anti-nutritional factors on key nutrients is established, and anti-nutritional interaction path data is generated. For example, a path can be established in which "trypsin inhibitors" reduce the digestibility of "crude protein" by inhibiting protease activity.

[0107] Step S4: extracting digestion influencing factors according to the anti-nutritional interaction pathway data to generate digestion influencing factor data; constructing a nutrient digestion prediction model according to the digestion influencing factor data to generate a nutrient digestion prediction model;

[0108] In an embodiment of the present invention, factors affecting the digestibility of nutrients are extracted based on the anti-nutritional interaction pathway data, and digestion influencing factor data is generated. For example, "trypsin inhibitor" concentration, "crude fiber" content, "fiber content", "enzyme activity" and the like can be extracted as digestion influencing factors. A nutrient digestion prediction model is constructed based on the digestion influencing factor data and the digestibility data of known forage nutrients using machine learning algorithms such as support vector machines, neural networks or random forests. A nutrient digestion prediction model is constructed using machine learning algorithms such as random forests, with digestion influencing factors as input variables and digestibility as output variables. During the model training process, a cross-validation method is used to evaluate the model performance, and the model parameters are optimized by methods such as grid search. For example, the digestibility of "crude protein" can be predicted based on factors such as "trypsin inhibitor" concentration and "crude fiber" content.

[0109] Step S5: using the nutrient digestion prediction model to predict the digestibility of the nutrient components of the acquired target forage sample, and generating the predicted digestibility of the nutrient components; performing a comprehensive nutritional value score based on the predicted digestibility of the nutrient components, and generating the nutritional value data of the forage;

[0110] In an embodiment of the present invention, first, a target forage sample is collected, and its component content is measured using a near-infrared spectrometer or other equipment, and then input into a terminal device to digitize it. Next, a weighted summation method is used to comprehensively score the predicted digestibility, and the weight can be determined according to the importance of different nutrients to animal growth. For example, the weight of crude protein digestibility can be set to 0.4, the weight of crude fiber digestibility can be set to 0.3, the weight of soluble sugar digestibility can be set to 0.2, and the weight of other trace element digestibility can be set to 0.1. The comprehensive nutritional value score of the forage is calculated by weighted summation.

[0111] Step S6: Optimizing the feeding formula according to the nutrient value data of the forage to generate optimized feeding plan data; evaluating the feeding effect according to the optimized feeding plan data to obtain feeding effect evaluation data.

[0112] In an embodiment of the present invention, according to the nutritional value data of forage, the feeding formula is optimized using optimization methods such as linear programming or mixed integer programming to generate optimized feeding scheme data. For example, according to the nutritional value data of different forages, the ratio of forages is optimized to meet the growth needs of animals, and the best feeding formula is generated. According to the optimized feeding scheme data, animal feeding experiments are carried out, and data such as animal growth indicators, blood indicators, and feces indicators are collected to evaluate the feeding effect and obtain feeding effect evaluation data. For example, the daily weight gain, feed conversion rate, blood biochemical indicators, etc. of animals under different feeding schemes can be compared to evaluate the effectiveness of the feeding scheme.

[0113] Preferably, step S1 comprises the following steps:

[0114] Step S11: Acquire forage literature resource data;

[0115] Step S12: performing data semantic cleaning on the forage literature resource data and performing text data deduplication processing to generate original forage corpus data;

[0116] Step S13: semantically annotating the original forage corpus data and performing named entity recognition to generate forage entity corpus data, wherein the forage entity data includes the forage name, nutritional components, anti-nutritional factors, and feeding animal species;

[0117] Step S14: performing ontology matching calculation on the forage entity corpus data and performing knowledge entity merging processing to generate forage entity data;

[0118] Step S15: extracting entity association relationships from the original forage corpus data through the forage entity data to generate entity association relationship data;

[0119] Step S16: constructing a knowledge graph based on the entity association relationship data and the forage entity data to generate an initial forage nutrition knowledge graph;

[0120] Step S17: performing forage knowledge link expansion processing on the forage literature resource data through the initial forage nutrition knowledge graph to obtain a forage nutrition knowledge graph.

[0121] In an embodiment of the present invention, a literature database such as PubMed, Web of Science or China Knowledge Network is used to search with keywords such as "forage", "nutrition", "ingredients", "digestibility", "anti-nutritional factors", "feeding animals", etc., to collect relevant literature resources. For example, the keyword "nutrient composition of forage" can be entered, and the search scope can be set to related fields such as "animal science" or "agricultural science" to obtain relevant documents. Natural language processing tools such as NLTK or spaCy are used for word segmentation, part-of-speech tagging and named entity recognition. Then, special characters, numbers and punctuation marks are removed by regular expression matching. For professional terms and abbreviations, a professional dictionary is established for unified processing, such as uniformly replacing "CP" with "crude protein". For example, brat can be used to annotate the original corpus data, identify entities such as "corn", "crude protein", "trypsin inhibitors", and "cows", and annotate the relationship between them, such as "corn contains crude protein", "trypsin inhibitors affect crude protein digestibility", and "cows eat corn". Named entity recognition tools such as Stanford NER or spaCy are used to perform named entity recognition on the annotated corpus data, extract entities such as forage names, nutrients, anti-nutritional factors, and feeding animal species, and generate forage entity corpus data. Semantic similarity algorithms such as the Jaccard coefficient are used to calculate the matching degree between entities and ontology concepts. For entities with a matching degree higher than a set threshold (such as 0.8), they are mapped to the corresponding ontology concepts. For entities with a lower matching degree, word vector models such as Word2Vec are used to calculate semantic similarity and find the closest ontology concept for matching. When merging, weighted voting is used to select the optimal attribute value, and the weight can be determined based on the reliability of the data source. For example, alfalfa and purple alfalfa are merged into a unified alfalfa entity. For example, OpenIE can be used to analyze the original corpus data to identify entity association relationships such as "corn contains crude protein", "trypsin inhibitors affect crude protein digestibility", and "cattle eat corn". The graph database Neo4j is used as a storage and query platform. The forage entity data is used as the nodes of the graph, and the entity association relationship data is used as the edges of the graph. For example, the "corn" entity, "crude protein" entity, "trypsin inhibitor" entity, "cattle" entity and the associations between them can be input into Neo4j to construct an initial forage nutrition knowledge graph. The initial forage nutrition knowledge graph is linked to other relevant open data sources (such as DBpedia, Wikidata), for example, the feed raw materials are linked to their corresponding Wikipedia pages, thereby enriching the information content of the knowledge graph.

[0122] Preferably, step S2 comprises the following steps:

[0123] Step S21: screening nutrient component nodes according to the forage nutrient knowledge graph to obtain forage nutrient component node data;

[0124] Step S22: performing co-occurrence frequency statistical processing according to the nutrient component node data of forage to generate nutrient co-occurrence frequency data;

[0125] Step S23: constructing a nutrient component co-occurrence matrix for the nutrient component node data of the forage by using the nutrient co-occurrence frequency data to generate a nutrient component co-occurrence matrix;

[0126] Step S24: performing singular value decomposition processing on the nutrient co-occurrence matrix and mining potential association relationships to generate nutrient potential association data;

[0127] Step S25: calculating the association strength of the potential association data of the nutrient components through the nutrient co-occurrence frequency data based on a preset support vector machine to obtain potential association strength data;

[0128] Step S26: inferring the type of association relationship based on the potential association strength data to generate nutrient component association type data;

[0129] Step S27: performing importance node evaluation on the forage nutrient component node data through the nutrient component association type data, and identifying key nutrient components to generate forage key nutrient component data.

[0130] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation steps of step S2 in the embodiment are shown in the flowchart. In this embodiment, step S2 includes:

[0131] Step S21: screening nutrient component nodes according to the forage nutrient knowledge graph to obtain forage nutrient component node data;

[0132] In an embodiment of the present invention, a knowledge graph visualization tool such as Neo4j Browser or Gephi is used to filter the forage nutrition knowledge graph according to the label of the "nutrient component" node, and the nodes related to the "nutrient component" such as "crude protein", "crude fat", "crude fiber", "ash", "water", etc. are filtered out to obtain the forage nutrition component node data.

[0133] Step S22: performing co-occurrence frequency statistical processing according to the nutrient component node data of forage to generate nutrient co-occurrence frequency data;

[0134] In the embodiment of the present invention, each forage entity in the knowledge graph is traversed according to the forage nutrient component node data, and the number of times the nutrient components appear at the same time is counted. For example, the nutrient component node data is grouped and counted, and the number of times each nutrient component appears at the same time with other nutrients is calculated and divided by the total number of documents to obtain the nutrient co-occurrence frequency data.

[0135] Step S23: constructing a nutrient component co-occurrence matrix for the nutrient component node data of the forage by using the nutrient co-occurrence frequency data to generate a nutrient component co-occurrence matrix;

[0136] In the embodiment of the present invention, all unique nutrients are determined and an n×n zero matrix is ​​created, where n is the number of nutrients. Then, the co-occurrence frequency data is traversed and the frequency values ​​are filled into the corresponding positions of the matrix. Each element in the matrix represents the frequency of co-occurrence of two nutrients, which can intuitively reflect the strength of the association between them. For example, if the co-occurrence frequency of protein and cellulose is high, the corresponding value in the matrix will be larger, which indicates that these two components often exist at the same time in the forage or have some functional association.

[0137] Step S24: performing singular value decomposition processing on the nutrient co-occurrence matrix and mining potential association relationships to generate nutrient potential association data;

[0138] In an embodiment of the present invention, the co-occurrence matrix is ​​decomposed into three matrices: U, Σ and V^T. Among them, U and V contain left singular vectors and right singular vectors, and Σ is a diagonal matrix containing singular values. Then, by analyzing the distribution of singular values, the number of principal components to be retained can be determined, and the scree plot method can be used to select the number of components before the inflection point of the singular value curve. Then, the low-dimensional matrix is ​​reconstructed using the retained principal components. This process can filter out noise and retain the most important association information. For example, if the first k principal components are retained, an approximate matrix can be obtained by matrix multiplication of U[:,:k], Σ[:k,:k] and V[:,:k]. Then, the element values ​​in the reconstructed matrix are analyzed, and larger values ​​indicate that there is a potential association relationship between the corresponding nutrient pairs.

[0139] Step S25: calculating the association strength of the potential association data of the nutrient components through the nutrient co-occurrence frequency data based on a preset support vector machine to obtain potential association strength data;

[0140] In an embodiment of the present invention, based on a preset support vector machine (SVM) model, the nutrient co-occurrence frequency data is used to calculate the association strength of the potential association data of the nutrient components. First, a feature vector is constructed, including features such as co-occurrence frequency and SVD reconstruction value. Then, the SVM model is trained using known strong and weak association samples. During the training process, the model parameters, such as the kernel function type and the penalty coefficient C, are optimized using grid search and cross-validation methods. Next, the trained SVM model is used to predict the potential association data, and the model will output a probability value between 0 and 1 to obtain the probability value of the association strength. For example, for the potential association pair "Vitamin A-β-carotene", the model can output a strong association probability of 0.85.

[0141] Step S26: inferring the type of association relationship based on the potential association strength data to generate nutrient component association type data;

[0142] In an embodiment of the present invention, first, a threshold of association strength is set, and association pairs are divided into two categories: strong association and weak association. For example, association pairs with an SVM predicted probability greater than 0.7 can be regarded as strong associations, those less than 0.3 can be regarded as weak associations, and those between the two can be regarded as medium associations. Then, for strong association pairs, their association types are further analyzed. For example, for the pair of associations "Vitamin A-β-carotene", if their co-occurrence frequency is high and their contents are positively correlated in a variety of forages, it can be inferred that they are a "precursor-product" relationship. For example: Complementary relationship: When two nutrients coexist, they can improve each other's absorption and utilization rate. Antagonistic relationship: When two nutrients coexist, they will reduce each other's absorption and utilization rate. Synergistic relationship: When two nutrients coexist, they will exert stronger physiological functions.

[0143] Step S27: performing importance node evaluation on the forage nutrient component node data through the nutrient component association type data, and identifying key nutrient components to generate forage key nutrient component data.

[0144] In an embodiment of the present invention, the forage nutrition knowledge map is analyzed. The NetworkX library is used to calculate the centrality index of each nutrient node, including degree centrality, betweenness centrality and eigenvector centrality. For example, for the "crude protein" node, a higher degree centrality value will be obtained, indicating that it is associated with multiple other nutrients. Combined with the nutrient association type data, a weight is assigned to each node. Strongly associated edges can be assigned higher weights, such as synergistic relationships can be assigned a weight of 1.5, and antagonistic relationships can be assigned a weight of 0.8. The importance score of the node is recalculated using the weighted PageRank algorithm. Set a threshold, such as selecting the top 20% of the nutrients as key ingredients. For example, for dairy cow feed, it will be found that crude protein, calcium, and phosphorus are the most critical nutrients.

[0145] Preferably, step S3 comprises the following steps:

[0146] Step S31: extracting anti-nutritional factors based on the key nutritional component data of forage through the forage nutrition knowledge map to generate anti-nutritional factor data;

[0147] Step S32: performing principal component coupling strength analysis on the key nutrient component data of forage through the anti-nutritional factor data to generate anti-nutritional factor-nutrient component coupling pair data;

[0148] Step S33: performing spectral clustering processing on the anti-nutrient factor-nutrient component coupling data, and performing action relationship pattern analysis to generate interaction relationship pattern data, wherein the interaction relationship pattern data includes synergistic action pattern, antagonistic action pattern and cumulative action pattern;

[0149] Step S34: conducting literature evidence mining on the interaction relationship pattern data, and quantifying the strength of the interaction relationship to generate interaction relationship strength data;

[0150] Step S35: Refining the interaction relationship type of the anti-nutritional factor-nutrient coupling data through the interaction relationship strength data, and constructing the interaction network to generate the anti-nutritional factor-nutrient interaction network;

[0151] Step S36: Identify anti-nutritional action relationship pathways based on the anti-nutritional factor-nutrient interaction network to generate anti-nutritional interaction pathway data.

[0152] As an example of the present invention, refer to Figure 3 As shown, Figure 1 Detailed implementation steps of step S3 in the flowchart, in this example, step S3 includes:

[0153] Step S31: extracting anti-nutritional factors based on the key nutritional component data of forage through the forage nutrition knowledge map to generate anti-nutritional factor data;

[0154] In an embodiment of the present invention, for example, using the Cypher query language of the Neo4j graph database, a query statement is written to identify anti-nutritional factor nodes directly connected to key nutrients. For example, for the key nutrient "lysine", related anti-nutritional factors such as "phytic acid" can be queried. Then, using a graph traversal algorithm such as depth-first search (DFS), the search range is expanded to second-degree or third-degree relationships to discover potential indirect anti-nutritional factors. Then, the extracted anti-nutritional factors are subjected to frequency statistics and importance ranking, and the TF-IDF algorithm can be used to calculate the weight of each factor. In addition, combined with text mining technology, anti-nutritional factors that are not clearly marked are extracted from the text attributes in the knowledge graph to form an anti-nutritional factor data set.

[0155] Step S32: performing principal component coupling strength analysis on the key nutrient component data of forage through the anti-nutritional factor data to generate anti-nutritional factor-nutrient component coupling pair data;

[0156] In the embodiment of the present invention, the principal component analysis method is used to perform coupling strength analysis on the anti-nutritional factor data and the key nutrient component data. For example, by analyzing the eigenvectors of the principal components, the anti-nutritional factor-nutrient component pairs that contribute the most to each principal component are identified. Next, the cosine similarity of each pair of factors in the principal component space is calculated as a quantitative indicator of their coupling strength. For example, if "tannin" and "protein" have similar loads in the main principal components, their coupling strength will be higher.

[0157] Step S33: performing spectral clustering processing on the anti-nutrient factor-nutrient component coupling data, and performing action relationship pattern analysis to generate interaction relationship pattern data, wherein the interaction relationship pattern data includes synergistic action pattern, antagonistic action pattern and cumulative action pattern;

[0158] In an embodiment of the present invention, a similarity matrix is ​​constructed based on coupling strength, and then the spectral clustering algorithm in the scikit-learn library of Python is used to cluster the coupling pairs. The optimal value of the number of clusters can be determined by the silhouette coefficient or the elbow method. For example, the coupling pairs are divided into 3-5 categories. Then, feature analysis is performed on each cluster to identify its representative action relationship pattern. A decision tree algorithm such as C4.5 can be used to extract the rules of each cluster, thereby summarizing the action mode. For example, if the coupling pairs in a cluster generally show positively correlated coupling strengths and there are records of synergy in the literature, they are classified as "synergistic action mode". Similarly, "antagonistic action mode" and "accumulative action mode" can be identified.

[0159] Step S34: conducting literature evidence mining on the interaction relationship pattern data, and quantifying the strength of the interaction relationship to generate interaction relationship strength data;

[0160] In an embodiment of the present invention, first, natural language processing techniques such as named entity recognition (NER) and relation extraction (RE) are used to retrieve relevant literature from academic databases and extract sentences describing the relationship. The extracted sentences are classified to identify supporting, opposing or neutral evidence. Each document is weighted, and the intensity of the effect of each antagonistic nutritional factor-nutrient and its confidence interval can be calculated based on factors such as publication time, number of citations and journal impact factor. For example, for the "phytic acid-zinc" relationship, it is concluded that phytic acid has a significant inhibitory effect on zinc absorption, with an average inhibition rate of 30% ± 5% (95% CI). Finally, information such as literature support, direction of action, intensity estimate and its uncertainty are integrated to form an interaction relationship strength data set.

[0161] Step S35: Refining the interaction relationship type of the anti-nutritional factor-nutrient coupling data through the interaction relationship strength data, and constructing the interaction network to generate the anti-nutritional factor-nutrient interaction network;

[0162] In an embodiment of the present invention, first, based on the interaction relationship strength data, a clustering algorithm such as K-means or hierarchical clustering is used to classify the interaction relationship more finely. For example, "antagonism" is further subdivided into "strong antagonism", "moderate antagonism" and "weak antagonism". Similarly, "synergy" can also be subdivided into different degrees. The topological properties of the network, such as degree centrality, betweenness centrality, etc., are calculated using the NetworkX library to identify key nodes and key relationships in the network. For example, it was found that "tannin" is a highly central anti-nutritional factor node that has antagonistic effects on a variety of amino acids. The importance is represented by the size of the node, the thickness of the edge represents the intensity of the action, and the color represents the type of action, so as to construct an anti-nutritional factor-nutrient interaction network.

[0163] Step S36: Identify anti-nutritional action relationship pathways based on the anti-nutritional factor-nutrient interaction network to generate anti-nutritional interaction pathway data.

[0164] In an embodiment of the present invention, for example, the constructed anti-nutritional factor-nutrient interaction network includes the following nodes and edges: nodes: trypsin inhibitor, lectin, crude protein, crude fiber; edges: trypsin inhibitor-crude protein (antagonistic effect), lectin-crude protein (antagonistic effect), lectin-crude fiber (synergistic effect). Use a graph algorithm such as the Dijkstra shortest path algorithm to find the path from the main anti-nutritional factors to the key nutrients in the interaction network. For example, trypsin inhibitor->crude protein: trypsin inhibitor reduces the digestibility of crude protein by inhibiting protease activity. Lectin->crude protein: lectins affect the absorption of crude protein by binding to intestinal epithelial cells.

[0165] Preferably, step S4 comprises the following steps:

[0166] Step S41: mining nutritional digestion knowledge based on the key nutritional component data of forage to generate nutritional digestion knowledge data;

[0167] Step S42: performing nutrient digestion characteristic analysis on the key nutrient component data of the forage using the nutrient digestion knowledge data to obtain nutrient digestion characteristic data;

[0168] Step S43: extracting digestion influencing factors from the nutrient digestion characteristic data based on the anti-nutritional interaction pathway data to generate digestion influencing factor data;

[0169] Step S44: performing digestion characteristic integration on the digestion influencing factor data and the nutrient digestion characteristic data to generate nutrient digestion characteristic data;

[0170] Step S45: constructing a nutrient digestion prediction model based on the nutrient digestion characteristic data to generate a nutrient digestion prediction model.

[0171] In the embodiment of the present invention, the forage nutrition knowledge map is used, and the key nutrient component data is used as the query condition to retrieve the digestion knowledge related to the key nutrient components, such as the digestibility of different nutrients of different animal species, the digestibility of different nutrients of different forage types, and the factors affecting the digestibility, etc., to generate the nutrition digestion knowledge data. For example, if the key nutrient component data includes "crude protein", the knowledge map can be used to query the digestion knowledge related to "crude protein", such as the digestibility of crude protein of ruminants is generally about 70%, while the digestibility of crude protein of monogastric animals is relatively low; the digestibility of crude protein in corn stalks is lower than that in alfalfa; anti-nutritional factors such as trypsin inhibitors will reduce the digestibility of crude protein, etc. The key nutrient component data of forage are analyzed, and the digestion characteristics of each key nutrient component are extracted, such as the digestibility range, the main factors affecting the digestibility, the digestibility differences under different animal species and forage types, etc., to generate the nutrient digestion characteristic data. For example, extract the digestion characteristics of "crude protein". For example, the digestibility of crude protein by ruminants generally ranges from 60% to 80%. The main influencing factors include the type of forage, the content of anti-nutritional factors, and the physiological condition of the animal. For example, based on the anti-nutritional action relationship path "trypsin inhibitor -> crude protein", "trypsin inhibitor concentration" can be extracted as a digestion influencing factor affecting the digestibility of crude protein. For example, "trypsin inhibitor concentration" is added to the digestion characteristic data of "crude protein" as a factor affecting the digestibility of crude protein. Using machine learning algorithms, such as linear regression, support vector machine, neural network, etc., a nutrient digestion prediction model is constructed based on the integrated nutrient digestion characteristic data. For example, a linear regression model can be used to establish a prediction model based on factors such as "trypsin inhibitor concentration", forage type, animal type, etc., to predict the digestibility of crude protein under different conditions.

[0172] Preferably, step S45 includes the following steps:

[0173] Step S451: Obtaining animal digestibility experimental data of forage;

[0174] Step S452: performing feature engineering processing on the experimental data of digestibility of forage animals to generate feature data of digestibility of forage animals;

[0175] Step S453: mapping the nutrient components and the forage digestibility using a random forest regression algorithm based on the forage animal digestibility characteristic data to construct a preliminary nutrient digestion prediction model;

[0176] Step S454: randomly dividing the nutrient digestion characteristic data into data sets to obtain a nutrient digestion training set and a nutrient digestion verification set;

[0177] Step S455: training the preliminary nutrition digestion prediction model using the nutrition digestion training set, and performing hyperparameter tuning according to the Michaelis equation to obtain an optimized nutrition digestion prediction model;

[0178] Step S456: Performing model verification processing on the optimized nutritional digestion prediction model through the nutritional digestion verification set to obtain the nutritional digestion prediction model.

[0179] In an embodiment of the present invention, experimental data on digestibility of forage animals are obtained from public databases, literature, research institutions and other channels. These data usually contain the following information: forage types: for example, corn stalks, alfalfa, soybean meal, etc.; forage ingredients: for example, crude protein, crude fat, crude fiber, ash and other contents; feeding animal types: for example, cattle, sheep, pigs, etc.; feeding methods: for example, free feeding, restricted feeding, etc.; digestibility index: for example, total digestibility, dry matter digestibility, organic matter digestibility, etc. The experimental data on digestibility of forage animals are subjected to feature engineering processing, including data cleaning, feature conversion, feature scaling, feature selection, etc. The random forest regression algorithm is used to map the nutrient component features in the digestibility feature data of forage animals generated in the digestibility index to the digestibility index, and a preliminary nutrient digestion prediction model is constructed. For example, a random forest model is initialized using the RandomForestRegressor class in the scikit-learn library of Python. Set initial parameters, such as n_estimators=100 (number of trees), max_depth=10 (maximum depth of the tree), min_samples_split=5 (minimum number of samples required for internal node re-division), etc. Then, divide the feature data into independent variable X (nutrient content) and dependent variable y (digestibility). Use the train_test_split function to divide the nutrient digestion feature data into a training set and a test set with a ratio of 8:2. Perform hyperparameter tuning based on the Michaelis-Menten equation. The Michaelis-Menten equation describes the relationship between the rate of an enzymatic reaction and the concentration of the substrate, which can be analogous to the relationship between model complexity and performance. For example, a callback function can be used to record the parameters and corresponding performance of each iteration, and a chart similar to the Michaelis-Menten curve can be drawn to intuitively show the relationship between model complexity and performance. According to the inflection point of the curve, the best parameter combination is selected. Hyperparameter tuning is performed according to the Michaelis-Menten equation to obtain an optimized nutrient digestion prediction model. The nutrient digestion validation set is used to validate the optimized nutrient digestion prediction model obtained through training. During the validation process, the data in the validation set is input into the model to predict the digestibility, which is compared with the actual digestibility to evaluate the prediction performance of the model.

[0180] Preferably, step S5 comprises the following steps:

[0181] Step S51: obtaining a target forage sample; performing feed sample informatization processing on the target forage sample to obtain digital forage sample data;

[0182] Step S52: using the forage nutrition knowledge map to match the digitized forage sample data with standard forage data to obtain matching forage nutrition data;

[0183] Step S53: performing sample-specific nutritional analysis according to the matched forage nutritional data to generate forage-specific nutritional data;

[0184] Step S54: performing environmental impact index analysis based on the forage-specific nutritional data to obtain a forage environmental adjustment coefficient;

[0185] Step S55: using a nutrient digestibility prediction model to predict nutrient digestibility of the digital forage sample data based on the forage environment adjustment coefficient, and generating a predicted nutrient digestibility;

[0186] Step S56: Calculate the digestible amount of nutrients according to the predicted digestibility of the nutrients to obtain the digestible amount data of the nutrients;

[0187] Step S57: Comprehensively score the nutritional value of the digitized forage sample data using the digestible amount data of the nutrient components to generate forage nutritional value data.

[0188] In an embodiment of the present invention, a target forage sample is obtained, and the target forage sample is collected from different sources, such as field sampling, market purchase or laboratory cultivation. Each sample is numbered and basic information is recorded, including type, collection location, collection time, etc. For example, a corn stalk sample used in actual production is collected. Then, the target forage sample is analyzed using a feed analysis method, such as near infrared spectroscopy analysis, chemical analysis, etc., to obtain its nutritional component data, such as crude protein content, crude fat content, crude fiber content, etc. Using the forage nutrition knowledge map, the nutritional component data in the digitized forage sample data is matched with the standard forage data stored in the knowledge map. For example, if the digitized forage sample data contains a crude protein content of 5% in corn stalks, the crude protein content of the standard corn stalks can be queried using the knowledge map and used as the matching forage nutrition data. For example, the difference between the crude protein content of the target corn stalk sample and the crude protein content of the standard corn stalk can be analyzed. Using a multiple regression analysis method, a relationship model between environmental factors and specific nutritional patterns is established. The linear regression model can be implemented using the statsmodels library. For each major nutrient (e.g. crude protein, crude fiber), a separate model was established. The influence of environmental factors on each nutrient was analyzed by the model coefficients. The variance inflation factor (VIF) was used to detect multicollinearity and ensure the stability of the model. Based on the regression model, the environmental impact index was calculated. For each sample, the theoretical deviation value of each nutrient was predicted according to its actual environmental conditions. The predicted value was compared with the actual observed specific deviation to obtain the environmental impact index. Based on the nutrient data in the digital forage sample data and the forage environmental adjustment coefficient, the digestibility of each nutrient in the target forage sample was predicted to generate the predicted digestibility of the nutrient. The two data sets were merged based on the sample ID using the merge function. Then, for each nutrient, its digestible amount was calculated. The digestible amount was calculated as follows: nutrient content multiplied by predicted digestibility. The confidence interval of the digestible amount of each nutrient was calculated. Assuming that the predicted digestibility follows a normal distribution, the 95% confidence interval can be calculated using the standard error. It can be obtained by weighted summing the digestible amount of each nutrient. The weights can be adjusted to suit the needs of different animals, for example, fiber will be weighted more heavily for ruminants. The content of each digestible nutrient is classified into four levels: excellent, good, fair, and poor. The weights reflect the relative importance of each nutrient, for example, protein will be weighted more heavily for dairy cows. Each sample is scored for each nutrient. For each nutrient, a score is assigned based on the level in which its digestibility falls, and then multiplied by the weight to get a weighted score. The total score is obtained by summing up the weighted scores of all nutrients.

[0189] Preferably, step S6 comprises the following steps:

[0190] Step S61: accurately dividing the growth stages of the target animals to obtain the growth stage data of the animals;

[0191] Step S62: quantifying the production performance target of the animal according to the growth stage data of the raised animal to obtain the production performance target data;

[0192] Step S63: Calculating the nutritional component requirement according to the production performance target data to generate animal nutritional component requirement data;

[0193] Step S64: optimizing the feeding formula of the target animal based on the nutrient value data of the forage and using the animal nutrient requirement data to generate optimized feeding plan data;

[0194] Step S65: formulating an accurate nutrition supply plan according to the optimized feeding plan data, and collecting feeding feedback data to obtain feeding feedback data;

[0195] Step S66: Perform feeding effect evaluation according to the feeding feedback data to obtain feeding effect evaluation data.

[0196] In an embodiment of the present invention, the entire growth process is divided into different stages according to the type, breed and growth and development law of the target animal. For example, for beef cattle, the growth stage can be divided into a fattening stage, a growth stage, a pregnancy stage, a lactation stage, etc. According to the growth stage data of the animal, combined with the production target, the animal production performance is quantified, such as daily weight gain, feed conversion rate, milk production, etc. According to the animal species, breed, growth stage and production target, the production performance targets of different stages are determined. According to the production performance target data, combined with animal nutrition knowledge, the requirements of animals for different nutrients at different growth stages are calculated, such as energy requirements, protein requirements, mineral requirements, etc. The existing nutritional requirement models, such as the NRC model and the ARC model, are used to calculate the nutritional requirements of the animal, and the requirements for different nutrients at each growth stage are recorded to form the animal nutritional requirement data. According to the animal nutritional requirement data and the nutritional value data of the forage, the ratio of the forage is optimized and the optimal feeding formula is generated using optimization methods such as linear programming or mixed integer programming. The optimization goal is to meet the nutritional needs of the animal and reduce the feed cost. Constraints such as feed type restrictions and feed cost restrictions can be set. According to the optimized feeding plan data and the actual situation of the animals, a precise nutrition supply plan is formulated, such as the daily feeding amount, feeding frequency, and the mixing ratio of forage. The feeding plan is adjusted according to the weight, age, and physiological state of the animals. For example, for animals with a certain weight, the feeding amount can be appropriately increased. During the feeding process, data such as the weight, feed intake, feces indicators, and blood indicators of the animals are collected as feeding feedback data. Using data analysis tools, the feeding effect is evaluated based on the collected feeding feedback data. For example, the daily weight gain, feed conversion rate, and blood biochemical indicators of the animals are analyzed to evaluate the effectiveness of the feeding plan. Different evaluation indicators are selected according to different production goals. For example, for beef cattle, daily weight gain, feed conversion rate, and slaughter rate can be evaluated. Statistical analysis methods, such as t-test and variance analysis, can be used to compare the production performance of animals under different feeding plans.

[0197] Preferably, step S64 includes the following steps:

[0198] Step S641: extracting the real-time forage supply of the target animal to obtain the real-time forage supply data;

[0199] Step S642: performing a nutrient demand gap analysis on the real-time forage supply data using the animal nutrient component demand data to obtain nutrient demand gap data;

[0200] Step S643: using the forage nutritional value data to match the nutritional demand gap data to compensate for the forage nutritional gap, and generating forage nutritional compensation data;

[0201] Step S644: Processing the availability of raw materials according to the nutrient compensation data of forage, and generating available feed raw material list data;

[0202] Step S645: generating a forage ratio scheme for the real-time forage supply data based on the available feed raw material list data to obtain forage ratio scheme data;

[0203] Step S646: Perform precise nutrient supply balance processing on the forage and grass ratio program data, and perform multi-objective optimization processing to generate optimized feeding program data.

[0204] In an embodiment of the present invention, a real-time forage supply extraction is performed on the target animal. For example, the sensor installed on the feeding equipment can record the animal's feed intake, feeding time and other information. Assuming that the target animal is a beef cattle, the sensor installed on the feeding equipment records that the beef cattle have eaten 5kg corn stalks and 2kg alfalfa in the past 1 hour. According to the real-time forage supply data, the amount of nutrients such as energy and protein currently ingested by the animal is calculated, and compared with the animal's nutritional requirements, the nutritional requirement gap is calculated. For example, assuming that the energy requirement of beef cattle is 10MJ / hour and the protein requirement is 500g / hour. According to the real-time forage supply data, the beef cattle have ingested 5kg corn stalks and 2kg alfalfa in the past 1 hour, and the energy intake is calculated to be 8MJ and the protein is 300g. Therefore, the current energy gap of the beef cattle is 2MJ and the protein gap is 200g. According to the gap in nutritional needs, choose forage that can effectively fill the gap. For example, if the energy gap is large, you can choose forage with higher energy content, for example, the energy content of corn is 10MJ / kg, and the protein content of soybean meal is 40%. According to the energy gap of 2MJ, it can be calculated that an additional 200g of corn is needed. According to the protein gap of 200g, it can be calculated that an additional 500g of soybean meal is needed. Evaluate whether the required forage can be obtained in time, for example, whether there is sufficient reserve and whether transportation is convenient. According to the available feed raw materials, calculate the best ratio to meet the nutritional needs of animals. For example, based on the available feed raw material list data, the following forage ratio can be generated: corn stalks: 50%, alfalfa: 30%, corn: 10%, soybean meal: 10%. According to the weight, age, physiological state, etc. of the animal, adjust the forage ratio to ensure that the animal can take in enough nutrition. The feeding plan can be optimized according to different goals, such as reducing feed costs, improving animal production performance, and reducing environmental impact. At the same time, the forage ratio plan can be optimized according to the feed cost, for example, reducing the proportion of soybean meal and increasing the proportion of corn straw to reduce feed costs.

[0205] The present invention also provides a forage nutrition knowledge graph construction and digestion analysis system, which executes the forage nutrition knowledge graph construction and digestion analysis method as described above, and the forage nutrition knowledge graph construction and digestion analysis system comprises:

[0206] A knowledge graph construction module is used to obtain forage literature resource data; and to construct a forage nutrition knowledge graph based on the forage literature resource data;

[0207] The nutrient component identification module is used to screen nutrient component nodes according to the forage nutrient knowledge map to obtain forage nutrient component node data; perform co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; perform key nutrient component identification according to the nutrient component co-occurrence matrix to generate forage key nutrient component data;

[0208] The anti-nutritional interaction analysis module is used to extract anti-nutritional factors based on the key nutritional component data of forage and generate anti-nutritional factor data; analyze the action relationship pattern of the key nutritional component data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identify the anti-nutritional action relationship path based on the interaction relationship pattern data to generate anti-nutritional interaction path data;

[0209] The digestion model construction module is used to extract digestion influencing factors based on the anti-nutritional interaction pathway data and generate digestion influencing factor data; construct a nutrient digestion prediction model based on the digestion influencing factor data and generate a nutrient digestion prediction model;

[0210] The nutritional value assessment module is used to predict the digestibility of the nutrient components of the acquired target forage samples using the nutritional digestion prediction model to generate the predicted digestibility of the nutrient components; perform a comprehensive nutritional value score based on the predicted digestibility of the nutrient components to generate the nutritional value data of the forage;

[0211] The intelligent feeding plan optimization module is used to optimize the feeding formula according to the nutritional value data of forage and generate optimized feeding plan data; and to evaluate the feeding effect according to the optimized feeding plan data to obtain the feeding effect evaluation data.

[0212] The beneficial effect of the present application is that, through the construction of the forage nutrition knowledge map, the scattered forage nutrition information is integrated into a structured knowledge network, so that the complex relationship between entities such as forage, nutrients, and anti-nutritional factors becomes clear and visible. By performing node screening and co-occurrence frequency analysis on the knowledge map, the data complexity can be effectively reduced, the most important nutritional elements can be highlighted, and the correlation between different nutrients can be revealed, and potential associations can be mined. Through the analysis of multiple nutritional component indicators, this provides a more comprehensive perspective for the evaluation of the nutritional value of forage. Taking the impact of anti-nutritional factors into consideration, the nutritional value of forage can be more comprehensively and objectively evaluated by identifying the anti-nutritional action path. By constructing a nutritional digestion prediction model, the digestibility of different forages in different animals can be more accurately predicted, overcoming the shortcomings of traditional methods that rely on empirical data and low prediction accuracy. The prediction results can help breeders avoid using feeds with low digestibility, thereby reducing feed waste and improving the nutrient absorption efficiency of animals. According to the nutritional needs of different animals, the nutritional value of the same forage is personalized and accurate feeding plans are formulated for different animals. The optimized formula can maximize the nutritional value of various forages while taking into account cost-effectiveness, providing animals with the best nutritional support, improving feed utilization efficiency, meeting the nutritional needs of animals, promoting healthy growth of animals, and improving animal production performance and product quality.

[0213] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0214] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for constructing a forage nutrition knowledge map and digestion analysis, characterized in that: The following steps are involved: Step S1: Acquire forage literature resource data; Construct a forage nutrition knowledge map based on forage literature resource data; Step S2: screening nutrient component nodes according to the forage nutrient knowledge graph to obtain forage nutrient component node data; Perform co-occurrence frequency statistics on the nutrient component node data of forage to generate a nutrient component co-occurrence matrix; Identify key nutrients based on the nutrient co-occurrence matrix and generate key nutrient data for forage; Step S3: extracting anti-nutritional factors according to the key nutritional component data of forage to generate anti-nutritional factor data; Through the anti-nutritional factor data, the interaction relationship model analysis is carried out on the key nutrient component data of forage to generate the interaction relationship model data; According to the interaction relationship pattern data, anti-nutritional action relationship path identification is performed to generate anti-nutritional interaction path data; wherein step S3 includes: Step S31: extracting anti-nutritional factors based on the key nutritional component data of forage through the forage nutrition knowledge map to generate anti-nutritional factor data; Step S32: performing principal component coupling strength analysis on the key nutrient component data of forage through the anti-nutritional factor data to generate anti-nutritional factor-nutrient component coupling pair data; Step S33: performing spectral clustering processing on the anti-nutrient factor-nutrient component coupling data, and performing action relationship pattern analysis to generate interaction relationship pattern data, wherein the interaction relationship pattern data includes synergistic action pattern, antagonistic action pattern and cumulative action pattern; Step S34: conducting literature evidence mining on the interaction relationship pattern data, and quantifying the strength of the interaction relationship to generate interaction relationship strength data; Step S35: Refining the interaction relationship type of the anti-nutritional factor-nutrient coupling data through the interaction relationship strength data, and constructing the interaction network to generate the anti-nutritional factor-nutrient interaction network; Step S36: identifying anti-nutritional action relationship pathways according to the anti-nutritional factor-nutrient interaction network, and generating anti-nutritional interaction pathway data; Step S4: extracting digestion influencing factors according to the anti-nutritional interaction pathway data to generate digestion influencing factor data; constructing a nutrient digestion prediction model according to the digestion influencing factor data to generate a nutrient digestion prediction model; Step S5: using the nutrient digestion prediction model to predict the digestibility of the nutrient components of the acquired target forage sample, and generating the predicted digestibility of the nutrient components; performing a comprehensive nutritional value score based on the predicted digestibility of the nutrient components, and generating the nutritional value data of the forage; Step S6: Optimizing the feeding formula according to the nutrient value data of the forage to generate optimized feeding plan data; evaluating the feeding effect according to the optimized feeding plan data to obtain feeding effect evaluation data.

2. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Acquire forage literature resource data; Step S12: performing data semantic cleaning on the forage literature resource data and performing text data deduplication processing to generate original forage corpus data; Step S13: semantically annotating the original forage corpus data and performing named entity recognition to generate forage entity corpus data, wherein the forage entity data includes the forage name, nutritional components, anti-nutritional factors, and feeding animal species; Step S14: performing ontology matching calculation on the forage entity corpus data and performing knowledge entity merging processing to generate forage entity data; Step S15: extracting entity association relationships from the original forage corpus data through the forage entity data to generate entity association relationship data; Step S16: constructing a knowledge graph based on the entity association relationship data and the forage entity data to generate an initial forage nutrition knowledge graph; Step S17: performing forage knowledge link expansion processing on the forage literature resource data through the initial forage nutrition knowledge graph to obtain a forage nutrition knowledge graph.

3. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: screening nutrient component nodes according to the forage nutrient knowledge graph to obtain forage nutrient component node data; Step S22: performing co-occurrence frequency statistical processing according to the nutrient component node data of forage to generate nutrient co-occurrence frequency data; Step S23: constructing a nutrient component co-occurrence matrix for the nutrient component node data of the forage by using the nutrient co-occurrence frequency data to generate a nutrient component co-occurrence matrix; Step S24: performing singular value decomposition processing on the nutrient co-occurrence matrix and mining potential association relationships to generate nutrient potential association data; Step S25: calculating the association strength of the potential association data of the nutrient components through the nutrient co-occurrence frequency data based on a preset support vector machine to obtain potential association strength data; Step S26: inferring the type of association relationship based on the potential association strength data to generate nutrient component association type data; Step S27: performing importance node evaluation on the forage nutrient component node data through the nutrient component association type data, and identifying key nutrient components to generate forage key nutrient component data.

4. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: mining nutritional digestion knowledge based on the key nutritional component data of forage to generate nutritional digestion knowledge data; Step S42: performing nutrient digestion characteristic analysis on the key nutrient component data of the forage using the nutrient digestion knowledge data to obtain nutrient digestion characteristic data; Step S43: extracting digestion influencing factors from the nutrient digestion characteristic data based on the anti-nutritional interaction pathway data to generate digestion influencing factor data; Step S44: performing digestion characteristic integration on the digestion influencing factor data and the nutrient digestion characteristic data to generate nutrient digestion characteristic data; Step S45: constructing a nutrient digestion prediction model based on the nutrient digestion characteristic data to generate a nutrient digestion prediction model.

5. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 4, characterized in that: Step S45 includes the following steps: Step S451: Obtaining experimental data on animal digestibility of forage; Step S452: performing feature engineering processing on the experimental data of digestibility of forage animals to generate feature data of digestibility of forage animals; Step S453: mapping the nutrient components and the forage digestibility using a random forest regression algorithm based on the forage animal digestibility characteristic data to construct a preliminary nutrient digestion prediction model; Step S454: randomly dividing the nutrient digestion characteristic data into data sets to obtain a nutrient digestion training set and a nutrient digestion verification set; Step S455: training the preliminary nutrition digestion prediction model using the nutrition digestion training set, and performing hyperparameter tuning according to the Michaelis equation to obtain an optimized nutrition digestion prediction model; Step S456: Performing model verification processing on the optimized nutritional digestion prediction model through the nutritional digestion verification set to obtain the nutritional digestion prediction model.

6. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 1, characterized in that: Step S5 includes the following steps: Step S51: obtaining a target forage sample; performing feed sample informatization processing on the target forage sample to obtain digital forage sample data; Step S52: using the forage nutrition knowledge map to match the digitized forage sample data with standard forage data to obtain matching forage nutrition data; Step S53: performing sample-specific nutritional analysis according to the matched forage nutritional data to generate forage-specific nutritional data; Step S54: performing environmental impact index analysis based on the forage-specific nutritional data to obtain a forage environmental adjustment coefficient; Step S55: using a nutrient digestibility prediction model to predict nutrient digestibility of the digital forage sample data based on the forage environment adjustment coefficient, and generating a predicted nutrient digestibility; Step S56: Calculate the digestible amount of nutrients according to the predicted digestibility of the nutrients to obtain the digestible amount data of the nutrients; Step S57: Comprehensively score the nutritional value of the digitized forage sample data using the nutrient digestible amount data to generate forage nutritional value data.

7. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 1, characterized in that: Step S6 includes the following steps: Step S61: accurately dividing the growth stages of the target animals to obtain the growth stage data of the animals; Step S62: quantifying the production performance target of the animal according to the growth stage data of the raised animal to obtain the production performance target data; Step S63: Calculating the nutritional component requirement according to the production performance target data to generate animal nutritional component requirement data; Step S64: optimizing the feeding formula of the target animal based on the nutrient value data of the forage and using the animal nutrient requirement data to generate optimized feeding plan data; Step S65: formulating an accurate nutrition supply plan according to the optimized feeding plan data, and collecting feeding feedback data to obtain feeding feedback data; Step S66: Perform feeding effect evaluation according to the feeding feedback data to obtain feeding effect evaluation data.

8. The method for constructing a forage nutrition knowledge map and digestion analysis according to claim 7, characterized in that: Step S64 includes the following steps: Step S641: extracting the real-time forage supply of the target animal to obtain the real-time forage supply data; Step S642: performing a nutrient demand gap analysis on the real-time forage supply data using the animal nutrient component demand data to obtain nutrient demand gap data; Step S643: using the forage nutritional value data to match the nutritional demand gap data to compensate for the forage nutritional gap, and generating forage nutritional compensation data; Step S644: Processing the availability of raw materials according to the nutrient compensation data of forage, and generating available feed raw material list data; Step S645: generating a forage ratio scheme for the real-time forage supply data based on the available feed raw material list data to obtain forage ratio scheme data; Step S646: Perform precise nutrient supply balance processing on the forage and grass ratio program data, and perform multi-objective optimization processing to generate optimized feeding program data.

9. A forage nutrition knowledge graph construction and digestion analysis system, characterized in that: Used to execute the forage nutrition knowledge graph construction and digestion analysis method as claimed in claim 1, the forage nutrition knowledge graph construction and digestion analysis system comprises: A knowledge graph construction module is used to obtain forage literature resource data; and to construct a forage nutrition knowledge graph based on the forage literature resource data; The nutrient component identification module is used to screen nutrient component nodes according to the forage nutrient knowledge map to obtain forage nutrient component node data; perform co-occurrence frequency statistical processing on the forage nutrient component node data to generate a nutrient component co-occurrence matrix; perform key nutrient component identification according to the nutrient component co-occurrence matrix to generate forage key nutrient component data; The anti-nutritional interaction analysis module is used to extract anti-nutritional factors based on the key nutritional component data of forage and generate anti-nutritional factor data; analyze the action relationship pattern of the key nutritional component data of forage through the anti-nutritional factor data to generate interaction relationship pattern data; identify the anti-nutritional action relationship path based on the interaction relationship pattern data to generate anti-nutritional interaction path data; The digestion model construction module is used to extract digestion influencing factors based on the anti-nutritional interaction pathway data and generate digestion influencing factor data; construct a nutrient digestion prediction model based on the digestion influencing factor data and generate a nutrient digestion prediction model; The nutritional value assessment module is used to predict the digestibility of the nutrient components of the acquired target forage samples using the nutritional digestion prediction model to generate the predicted digestibility of the nutrient components; perform a comprehensive nutritional value score based on the predicted digestibility of the nutrient components to generate the nutritional value data of the forage; The intelligent feeding plan optimization module is used to optimize the feeding formula according to the nutritional value data of forage and generate optimized feeding plan data; and to evaluate the feeding effect according to the optimized feeding plan data to obtain the feeding effect evaluation data.

Images