A method and system for automatically recommending crop breeding materials

Abstract

A method and system for automatically promoting and recommending crop breeding materials includes the following steps: S1: obtaining phenotypic data of the breeding materials; S2: obtaining trait index data; S3: creating a promotion recommendation model, which allows the user to customize promotion and recommendation conditions; the system simultaneously combines the trait index data within the system into a training dataset, optimizes the promotion and recommendation conditions, and obtains the optimized promotion recommendation model through verification with a test dataset; S4: selecting the promotion recommendation model to process and analyze the trait index data of the breeding materials, obtains promotion coefficients and recommendation coefficients, and performs recommendation rankings; S5: displaying the analysis results. The present invention automatically promotes and recommends high-quality materials by simply inputting test data, eliminating the need for breeders to rely on experience to check each item individually or to organize and compile data using spreadsheets. This method is highly efficient, allowing the system to quickly screen out materials with the best comprehensive indicators from a large number of indicators and recommend them to breeders.

Description

Technical Field

[0001] The present invention relates to the cross-technical field of agricultural breeding and computer informationization, and in particular to a method and system for automatically promoting and recommending crop breeding materials. Background Art

[0002] In the current breeding process, selecting high-quality materials is essentially a manual process of checking various breeding indicators. Breeders then use their experience to select suitable breeding materials for subsequent breeding, or to select high-quality crop varieties for commercial development. Some breeders also use spreadsheet tools such as Excel to sort and filter data. Current technology has the following shortcomings:

[0003] 1. Over-reliance on the experience of breeders and checking and analyzing data one by one is time-consuming, labor-intensive, and inefficient. Moreover, for those who lack experience, it is difficult to quickly select the optimal comprehensive indicator from a large number of indicators.

[0004] 2. The analysis of breeding data requires simultaneous analysis of multiple indicators and parameters. Using Excel for simple sorting and filtering cannot quickly integrate multiple indicators for analysis. Organizing data and writing formulas each time is also time-consuming and cannot achieve data integration and information processing.

[0005] 3. When the data is incomplete, for example, three years of test data are required but only two years have been tested, if it is processed manually, a large amount of data needs to be manually supplemented, which is very cumbersome; if it is processed using Excel, Excel cannot automatically supplement the data and can only eliminate the materials; this can neither achieve efficient data processing nor ensure that as much high-quality materials as possible are retained.

[0006] 4. Existing agricultural breeding technologies that combine computer science mostly analyze at least one phenotypic trait of a crop, then improve that phenotypic trait and select high-quality crop varieties through breeding methods. However, this technology generates a large number of candidate materials during implementation. Current technologies cannot automatically recommend high-quality materials that meet the requirements from this large number of candidate materials, and therefore cannot solve the technical problem of the present invention. Summary of the Invention

[0007] The purpose of the present invention is to overcome the above-mentioned deficiencies in the prior art and to provide a method and system for automatically recommending crop breeding materials with high automation, high efficiency, precise screening and simplified calculation.

[0008] The technical solution of the present invention is:

[0009] A method for automatically recommending crop breeding materials for advancement according to the present invention comprises the following steps:

[0010] S1: Obtain phenotypic data of breeding materials;

[0011] S2: performing trait index analysis on the acquired phenotypic data to obtain trait index data;

[0012] S3: Create a promotion recommendation model, which allows users to customize promotion and recommendation conditions. The system then uses the phenotypic indicator analysis data within the system to form a training dataset, optimizes the promotion and recommendation conditions using the training dataset, and verifies the optimized promotion recommendation model using the test dataset.

[0013] S4: Select the promotion recommendation model to process and analyze the breeding material trait index data. If the breeding material does not meet the promotion conditions, it will be eliminated; if it meets the promotion conditions, the material will be promoted and obtain the promotion coefficient for ranking display;

[0014] S5: Conduct recommendation analysis on the breeding materials that advance according to the recommendation conditions, calculate the recommendation coefficient, and recommend ranking the breeding materials that advance according to the recommendation coefficient;

[0015] S6: Display analysis results.

[0016] Furthermore, the promotion recommendation model includes promotion conditions and recommendation conditions; first, it is determined whether each trait index of the breeding material meets the promotion conditions, and the promotion coefficient is calculated; if not, the breeding material is considered to be eliminated; if it meets, it is analyzed according to the recommendation conditions and the recommendation coefficient is calculated;

[0017] The score is calculated based on the number of promotion conditions and recommendation conditions met by the breeding materials. One point is added for each item that is met, and each item is assigned a corresponding weight to obtain the promotion coefficient or recommendation coefficient. Similarly, the scores of all promotion conditions or recommendation conditions are counted to summarize the promotion coefficient or recommendation coefficient. All breeding materials are ranked from high to low according to the recommendation coefficient. Based on the ranking, excellent materials are recommended to users.

[0018] Furthermore, the optimized promotion recommendation model includes optimizing promotion conditions, and the method for obtaining the optimized promotion conditions includes:

[0019] Obtain historical phenotypic data of breeding materials, perform trait index analysis on the historical data, organize the analyzed data into a promotion training data set, calculate the trait threshold of each trait index, and use the trait threshold as a candidate promotion condition for testing the test data set; manually set the promotion pass ratio of the test data set, and when the test data set exceeds this ratio, it will be promoted, and the candidate promotion condition will be used as the optimized promotion condition; if the number of promoted materials in the test data set does not reach this ratio, readjust the corresponding trait threshold range, and continue testing the test data set after adjustment until the optimized promotion condition is obtained.

[0020] Furthermore, the optimized promotion recommendation model also includes optimized recommendation conditions, and the method for obtaining the optimized recommendation conditions includes:

[0021] Obtain historical phenotypic data of breeding materials, perform trait index analysis on the historical data, organize the analyzed data into a recommended training data set, calculate the trait threshold of each trait index, and use the trait threshold as a candidate recommendation condition for testing the test data set; manually set the recommended ranking pass ratio of the test data set, and when the test data set exceeds this ratio, it will be promoted, and the candidate recommendation condition will be used as the optimized recommendation condition; if the number of recommended materials in the test data set does not reach this ratio, readjust the corresponding trait threshold range, and continue testing the test data set after adjustment until the optimized recommendation condition is obtained.

[0022] Furthermore, the calculation method of the promotion coefficient or recommendation coefficient in the promotion recommendation model includes:

[0023] The trait index data is converted into a decimal greater than 0 and less than 1, and then added with the "1" point representing that the conditions are met, and then multiplied by the weight to serve as the promotion coefficient or recommendation coefficient; or the converted decimal is first multiplied by the weight, and then "1" is added as the promotion coefficient or recommendation coefficient; when the value of a certain trait index is as small as possible, first convert the value into a decimal greater than 0 and less than 1, and then subtract the decimal from 1, and then add with the "1" point representing that the conditions are met, multiplied by the weight, or first multiplied by the weight, and then "1" is added as the promotion coefficient or recommendation coefficient.

[0024] Furthermore, in the promotion recommendation model, there is also an option of forced passing; even if the breeding material does not meet the conditions,

[0025] Elimination: Users can make forced passes, forced promotions or forced recommendations based on their needs.

[0026] Furthermore, the trait indicator data includes one or more of average value, estimated value, control percentage, breeding value, ranking, breeding value ranking, minimum value, maximum value, adjusted estimated value, least squares mean, standard deviation, stability, estimated value ranking, significance, F value, variance, number of yield increase points, and proportion of yield increase points.

[0027] The present invention provides an automatic promotion and recommendation system for crop breeding materials, comprising:

[0028] A data acquisition module is used to obtain phenotypic data of breeding materials;

[0029] A data analysis module is used to calculate the corresponding trait indicators of the phenotypic data;

[0030] A data storage module, used to store phenotypic data and trait index data analyzed by the data analysis module;

[0031] The promotion recommendation management module includes a promotion recommendation model management module and a promotion recommendation analysis task management module; the promotion recommendation model management module is used to create a promotion recommendation model; the promotion analysis task management module is used to create a promotion recommendation analysis task;

[0032] The rule engine processor is used to take the trait index data separated by the data analysis module as input, enter different rule models for operation, calculate the promotion coefficient and recommendation coefficient, and count the scores of all promotion conditions or recommendation conditions. It also ranks all breeding materials from high to low according to the recommendation coefficient;

[0033] The result display module is used to display the analysis results.

[0034] Furthermore, the rule engine processor includes a rule set area, a calculation area, and a data area;

[0035] The rule set area contains multiple rules that are set, and each rule contains specific business rule calculation logic; the rules refer to the trait indicator data range of the achievable promotion conditions and recommendation conditions set in the promotion recommendation model;

[0036] The calculation area is used to calculate, summarize and rank the promotion coefficients and recommendation coefficients in the promotion recommendation model;

[0037] The data area is used to store the data of each rule, mainly including rule parameters, calculation results and rule weights.

[0038] Furthermore, it also includes an optimization and promotion recommendation model module, which includes:

[0039] Model optimization analysis module, used for model optimization, including optimization of promotion conditions and recommendation conditions;

[0040] The training data set is used to store training data integrated with historical data of the model;

[0041] The test data set is used to obtain promoted and recommended breeding materials, test the preliminary optimized model, test whether the optimized model is suitable, and provide appropriate optimized promotion conditions and optimized recommendation conditions.

[0042] Beneficial effects of the present invention:

[0043] (1) On the one hand, only the test data needs to be input to automatically recommend excellent materials and display the details of the promotion recommendation, which saves time and effort. There is no need for breeders to rely on experience to check each item one by one, nor is there a need to use Excel to organize data and write formulas, which greatly improves work efficiency. Thus, among a large number of indicators, the materials with the best comprehensive indicators can be quickly screened out and recommended to breeders; on the other hand, when the data is incomplete, only new data needs to be imported, and the system will automatically integrate and analyze the historical data and new data. The operation is very simple, so there will be no various errors like manual supplementation, thereby ensuring that as many high-quality materials as possible are retained, and accurate screening can be achieved through efficient data processing.

[0044] (2) The recommendation coefficient is obtained by adopting weights. Since the influence proportional factors of different trait indicators in calculating the promotion results are different, the concept of weights is introduced mainly as a means of differentiation. The weights currently need to be combined with the experience of breeders and different weights can be customized according to the conditions.

[0045] (3) By setting up a data analysis module to calculate the trait indicators of phenotypic data, and then sending the indicator data to the rule engine processor for calculation and analysis, loose coupling can be achieved. That is, the trait indicators are data closely related to the business, and the rule engine itself does not have any business logic attributes. After the indicator data is calculated, it is input into the rule engine, which can well separate the business logic and application logic. If the promotion logic adjustment is involved in the later stage, it is only necessary to modify the corresponding rule set to meet the needs.

[0046] (4) After the phenotypic data within the system are analyzed by indicators, a training data set is formed. The thresholds of various trait indicators are automatically calculated through the training data set, and the promotion recommendation model is automatically predicted for breeders. At the same time, the data of the promotion recommendation materials are used as a test data set to automatically judge and adjust the promotion conditions and recommendation conditions of the predicted promotion recommendation model, realizing the function of automatically optimizing the model using the data within the system, and using the new promotion and recommendation conditions for future breeding selection, supporting breeders to screen breeding materials with more stringent conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 is a schematic diagram of the process of Example 1 of the present invention;

[0048] Figure 2 2 is a schematic diagram of the promotion recommendation model analysis interface of Example 1 of the present invention;

[0049] Figure 3 yes Figure 2 The left side of the promotion recommendation model analysis interface of Example 1 is shown as an enlarged schematic diagram;

[0050] Figure 4 yes Figure 2 The right side of the promotion recommendation model analysis interface of Example 1 is shown as an enlarged schematic diagram;

[0051] Figure 5 Schematic diagram of the promotion model analysis results and recommended coefficients of Example 1 of the present invention;

[0052] Figure 6 This is the promotion recommendation result display interface of Example 1 of the present invention;

[0053] Figure 7 This is the promotion recommendation result confirmation interface of Example 1 of the present invention;

[0054] Figure 8 This is a schematic diagram of the optimization process of the promotion recommendation model according to Example 2 of the present invention;

[0055] Figure 9 This is a schematic diagram of the interface for creating a model optimization task in Example 2 of the present invention;

[0056] Figure 10 2 is a schematic diagram showing the optimization results after the model is optimized in Example 2 of the present invention;

[0057] Figure 11 This is a partial screenshot of the training data set of Example 2 of the present invention;

[0058] Figure 12 This is a partial screenshot of the test data set of Example 2 of the present invention;

[0059] Figure 13 This is the data that was re-upgraded after the model optimization of Example 2 of the present invention;

[0060] Figure 14 This is the data recommended again after the model optimization of Example 2 of the present invention. DETAILED DESCRIPTION

[0061] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0062] Example 1

[0063] A crop breeding material automatic promotion recommendation system, comprising:

[0064] (1) Data acquisition module, used to obtain phenotypic data of breeding materials. Specifically, the data acquisition methods mainly include: Method 1: Use a mobile phone app to record field phenotypic data. After recording, click the data synchronization function button on the app to synchronize the data to the system's data storage module; Method 2: Use a spreadsheet to directly import field phenotypic data into the system's data storage module. Among them, phenotypic data are data filled in for various traits of breeding materials, such as per-acre yield, plant height, flowering period, disease resistance, etc.

[0065] Among them, the breeding materials of this embodiment also include crop varieties, that is, they include both materials produced during the breeding process and crop varieties that have official identities after approval.

[0066] (2) Data analysis module, used to calculate the corresponding trait indicators of phenotypic data, including but not limited to: average value, estimated value, control percentage, breeding value, ranking, breeding value ranking, minimum value, maximum value, adjusted estimated value, least square mean, standard deviation, stability, estimated value ranking, significance, F value, variance, number of yield increase points, proportion of yield increase points, etc. For example: the obtained phenotypic data are grouped, and each group is subjected to the above-mentioned index calculation by the data analysis module to obtain all trait indicators. For example, the relevant data of plant height obtained are calculated by the data analysis module to obtain the above-mentioned average value, estimated value, etc. Among them, the above-mentioned "ranking" refers to the ranking of the traits of a group of data according to their quality. It is understandable that the trait indicators can be selected according to needs, or all can be selected by default.

[0067] (3) Data storage module, used to store phenotypic data and indicator data analyzed by the data analysis module, so as to facilitate subsequent phenotypic promotion analysis. Specifically, the data storage module uses a relational database.

[0068] (4) Promotion recommendation management module, including promotion recommendation model management module and promotion recommendation analysis task management module.

[0069] The promotion recommendation model management module is used to create promotion recommendation models and edit, delete, and query existing promotion recommendation models. Promotion recommendation models are used in subsequent promotion recommendation analysis. The promotion conditions and recommendation conditions in the promotion recommendation model serve as the basis for promotion analysis, promotion evaluation, recommendation analysis, and recommendation evaluation.

[0070] The promotion recommendation analysis task management module is used to create new promotion analysis tasks to analyze tasks with different needs; completed promotion tasks will be stored in the system, and users can view, delete, and query completed tasks.

[0071] (5) Optimize the promotion recommendation model module, which is used to optimize the promotion recommendation model, including the model optimization analysis module, training data set and test data set.

[0072] The model optimization analysis module is used for model optimization, including optimization of promotion conditions and optimization of recommendation conditions. After the phenotypic data is analyzed by indicators, it forms a training data set. The initial optimization of the model is based on the analysis of the training data set. The test data set is used to obtain promotion and recommended breeding materials, test the initial optimized model, and test whether the optimized model is suitable. For the specific optimization process of the model, please refer to Figure 8 .

[0073] The specific process for optimizing promotion conditions is as follows: Historical data is synthesized and analyzed for trait indicators of breeding materials. This analyzed data is then combined into a promotion training dataset for optimizing promotion conditions. The data for all promoted materials is then combined to form a promotion test dataset.

[0074] The threshold value of each trait is calculated through BioCloud's analysis model. The calculation method includes but is not limited to: for a single trait, the threshold value needs to achieve a 5% material pass rate. The calculated trait threshold is used as a candidate promotion condition for testing the test data set. The pass ratio of the test data set is manually set, for example, to 95%. When more than 95% of the materials in the test data set can be promoted, the candidate promotion condition can be saved as a recommended promotion condition; when the number of promoted materials in the test data set does not reach 95%, the threshold range is readjusted, and a new threshold value is recalculated based on BioCloud's analysis model. The calculation method includes but is not limited to: based on the distribution of material data, readjust the threshold value of one or several traits, continue to test the test data set after adjustment, and through iterative calculation, finally save the promotion condition that meets the test requirements as the optimized condition.

[0075] The specific process for optimizing recommended conditions is as follows: Historical data is synthesized and analyzed for trait indicators of breeding materials. This analyzed data is then used to form a training dataset for optimizing recommended conditions. The data for all recommended materials is then combined to form a recommended test dataset.

[0076] The threshold value of each trait is calculated through BioCloud's analysis model. The calculation method includes but is not limited to: for a single trait, the threshold value needs to achieve a 5% material pass rate, or a recommended ranking is given based on the pass rate. The calculated trait threshold is used as a candidate recommendation condition for testing the test data set. The pass rate of the test data set is manually set, for example, to 95%. When more than 95% of the materials in the test data set can be recommended, the candidate recommendation condition can be saved as the recommended recommendation condition; when the number of recommended materials in the test data set does not reach 95%, the threshold range is readjusted, and a new threshold is recalculated based on BioCloud's analysis model. The calculation method includes but is not limited to: based on the distribution of material data, readjust the threshold value of one or several traits. After adjustment, the test data set is continued to be tested, and through iterative calculation, the recommended condition that finally meets the test requirements is saved as the optimized condition.

[0077] (6) Rule engine processor, including rule set area, calculation area and data area.

[0078] The rule set area contains multiple rules, each of which contains specific business rule calculation logic. Rules refer to the achievable promotion and recommendation criteria set in the promotion recommendation model. The rule engine processor uses the trait indicator data analyzed by the data analysis module as input and sequentially applies it to different rule models for calculation. Together, these models drive the accumulation of promotion and recommendation coefficients.

[0079] The calculation area is used to calculate, summarize and rank the promotion coefficients and recommendation coefficients in the promotion recommendation model. For example, in the promotion recommendation model, the promotion status of the material is first analyzed, that is, the trait index data input by the data analysis module is calculated to obtain the promotion coefficient. The calculation of the promotion coefficient includes: every time a rule is passed in the rule set area, the result of the calculation area is automatically accumulated and added by 1; preferably, the trait index data is first converted into a numerical value of the same type, such as a decimal, and then multiplied by the corresponding weight coefficient, and finally the result is added by 1, which is the promotion coefficient. Allow users to set different levels of importance for different conditions, and set different weight coefficients according to the importance of the screening conditions. If the user believes that the importance of all promotion conditions is the same, then the weight of each condition is the same, so the weight setting can be omitted. For promoted materials, it is necessary to use the recommendation conditions in the promotion recommendation model to recommend the materials. Recommendation analysis calculates the trait index data input by the data analysis module, and whenever a rule is passed in the rule set area, the result of the calculation area is automatically accumulated and added by 1; preferably, the trait index data is first converted into a numerical value of the same type, such as a decimal, and then multiplied by the corresponding weight coefficient, and finally the result is added by 1, which is the recommendation coefficient. Users are allowed to set different levels of importance for different conditions, and set different weight coefficients according to the importance of the screening conditions. If the user believes that the importance of all promotion conditions is the same, then the weight of each condition is the same, and the weight setting can be omitted. The free setting of weight coefficients makes it easier for breeders to choose more suitable materials according to different breeding purposes.

[0080] The data area is the data storage area for each rule, which mainly includes rule parameters, calculation results, and rule weights.

[0081] (7) Result display module: uses visual display methods to present the analysis results to customers, including data table display, result graphic display, etc.

[0082] like Figure 1 As shown: A method for automatically promoting and recommending crop breeding materials, comprising the following steps:

[0083] S101: Obtain phenotypic data of breeding materials.

[0084] Specifically, the user needs to obtain the phenotypic data of the breeding material. The phenotypic data can be new data imported using the data acquisition module or historical data in the promotion system. After the phenotypic data is obtained, it is saved in the data storage module.

[0085] S102: Performing index analysis on the acquired phenotypic data to obtain trait index data.

[0086] Specifically, the data analysis module calculates corresponding indicators of the phenotypic data, including but not limited to: average value, estimated value, control percentage, breeding value, ranking, breeding value ranking, minimum value, maximum value, adjusted estimated value, least squares mean, standard deviation, stability, estimated value ranking, significance, F value, variance, number of yield-increasing points, and proportion of yield-increasing points. In the embodiments, the average value is used as an example, and the same applies to other trait indicators.

[0087] S103: Based on the trait index data, a promotion recommendation model is selected to process and analyze the trait index data. If the breeding material does not meet the promotion criteria, it is eliminated. If the promotion criteria are met, a promotion coefficient is assigned, and the selected breeding material is further analyzed according to the recommendation criteria, and a recommendation coefficient is assigned to the material. If the recommendation criteria are met, all breeding materials are ranked by calculating the recommendation coefficient, and excellent breeding materials are recommended to the user based on the ranking.

[0088] Specifically, a promotion recommendation model is created through the promotion recommendation model management module; corresponding rules are set in each model through the rule engine processor; the trait indicator data analyzed by the data analysis module is used as input, and enters different rule models in turn for calculation, and the promotion coefficient and recommendation coefficient are summarized.

[0089] Among them, the promotion recommendation model consists of two parts: the first part is the promotion conditions, and the second part is the recommendation conditions.

[0090] Promotion conditions refer to the conditions that breeding materials must meet (i.e. rules, such as national standards, etc.). If the requirements are not met, the materials will be judged to be eliminated.

[0091] After eliminating the breeding materials that do not meet the conditions through the promotion conditions, the remaining materials are given promotion coefficients, and analyzed according to the recommended conditions to recommend excellent materials to users. The specific analysis method of the recommended conditions is:

[0092] Convert the trait index data of a certain breeding material into a decimal greater than 0 and less than 1, add it to the score of "1" representing that the conditions are met, and then multiply it by the weight to use it as the recommendation coefficient; or multiply the converted decimal by the weight first, and then add "1" to use it as the recommendation coefficient.

[0093] Furthermore, when the smaller the value of a certain trait indicator, the better, first convert the value into a decimal greater than 0 and less than 1, then subtract the decimal from 1, and then add it to the "1" point representing promotion and multiply it by the weight, or first multiply it by the weight and then add "1" point to serve as the recommendation coefficient.

[0094] Similarly, the scores of all recommended conditions are counted and the recommendation coefficient is summarized. All breeding materials are ranked from high to low according to the recommendation coefficient. The system will recommend high-quality materials to users based on the ranking and inform users of the basis for promotion.

[0095] At the same time, this embodiment supports users (breeders) to set different weights for various traits of promotion conditions or recommendation conditions. Since breeders have accumulated many years of breeding experience, have a deep understanding of breeding, and have a stronger understanding of their own breeding goals, the system allows users to set different weights for different promotion or recommendation conditions. In this way, more distinctive candidate materials can be selected to achieve different screening purposes.

[0096] In this embodiment, the system allows adding new conditions or changing corresponding weights to the promotion conditions or recommendation conditions of any model to form a new model; based on the new model, personalized variety screening is completed.

[0097] It is understood that for breeding materials that are eliminated under the promotion conditions, the promotion coefficient can also be calculated and marked to distinguish them from the promotion coefficients of the promoted breeding materials, for example, by adding an "*" to the promotion coefficient. The promotion coefficient is also calculated for the eliminated breeding materials for reference by users. If the user believes that the material can be promoted, the user can force the material to be approved, so that the user can retain certain materials of special value.

[0098] In this embodiment, the required promotion recommendation model can be selected in the system operation interface to create an analysis task.

[0099] The following are preferred embodiments of the present invention for completing the breeding material promotion recommendation analysis based on different promotion recommendation models:

[0100] The rule model in this embodiment mainly includes the following three types of rule judgment criteria for trait index data:

[0101] (1) Criteria for judging numerical data

[0102] For example, if the average yield per mu is greater than 600 kg / mu in the rule model, the candidate material's average yield per mu value needs to be compared with 600. If the candidate material's average yield per mu is 700 kg / mu, then 700>600, and it passes. If the candidate material's average yield per mu is 599, and 599<600, it is eliminated.

[0103] Specifically, when calculating promotion or recommendation coefficients for numerical data, a conversion factor is set within the system based on the data's normal range. This involves first converting the data to a decimal, adding the value "1" to represent satisfied conditions, and then multiplying it by a weight to determine the promotion or recommendation coefficient. For example, if the production numbers of several materials are 500, 600, 700, and 750, respectively, the highest value, 750, is used as the basis, as 750 is the hundreds digit. Then, thousands is used as the divisor to convert the value to a decimal not greater than 1. This value is then added to the value "1" to represent satisfied conditions, and then multiplied by the weight to determine the aggregated promotion or recommendation coefficient for analysis.

[0104] For another example, under the trait of "yield", the default weight is "1", material A is "710", material B is "690", and the promotion condition is 600. Both A and B meet the promotion conditions. Then, under the trait of "yield" (the default weight is "1"), the promotion coefficient obtained by A is (1+710 / 1000)*1=1.71, and the promotion coefficient obtained by B is (1+690 / 1000)*1=1.69. Therefore, A obtains a higher promotion coefficient than B.

[0105] In particular, if the value of a trait is considered as small as possible, you need to first subtract the value from 1, then add it to the "1" representing the condition being met, and then multiply it by the weight to use as the promotion coefficient or recommendation coefficient. For example, under the "Rice Blast Loss Rate" trait, the default weight is "1", the A material is "5%", the B material is "7%", and the promotion condition is 10%. If both A and B meet the promotion conditions, then under the "Rice Blast Loss Rate" trait (the default weight is "1"), the promotion coefficient obtained by A is [1 + (1-5%)] * 1 = 1.95, and the promotion coefficient obtained by B is [1 + (1-7%)] * 1 = 1.93. Therefore, A receives a higher promotion coefficient than B.

[0106] (2) Judgment criteria for tabular data

[0107] For example, if a tabular value appears in the promotion recommendation model, such as the rice blast resistance level, which is divided into high resistance, resistance, moderate resistance, susceptible, moderately susceptible, and highly susceptible, the system will automatically convert the tabular text into numbers based on the information defined by the disease resistance level. After conversion to numbers, the numerical values ​​can be compared. An example of the conversion rules is shown in Table 1:

[0108] Table 1

[0109]

[0110] In particular, referring to Table 1, when calculating the promotion coefficient or recommendation coefficient for tabular data, the numerical value of the tabular text converted into a number is defined as a, a<10. The numerical value a is first converted into a decimal, that is, converted into a number greater than 0 and less than 1, for example, calculated according to (10-a) / 10. After conversion to a decimal, it is added to "1" representing that the conditions are met, and the result is multiplied by the weight to obtain the promotion coefficient or recommendation coefficient.

[0111] For example, for the trait "Blast Resistance (Grade)" with a default weight of "1," material A is "Resistant," and after conversion, it becomes "1." Material B is "Medium Resistance," and after conversion, it becomes "3." If both A and B meet the promotion criteria, then under the trait "Blast Resistance (Grade)" (default weight of "1"), A's promotion coefficient is [1 + (10-1) / 10] *1 = 1.9, and B's promotion coefficient is [1 + (10-3) / 10] *1 = 1.7. Therefore, A receives a higher promotion coefficient than B.

[0112] (3) Date type data judgment criteria

[0113] For date type data, comparison is performed in the order of dates, for example, May 2, 2022 > May 1, 2022.

[0114] In summary, the three data types described above are independently determined. They appear in the selection criteria used by breeders during the breeding process. For example, average yield per mu (700 kg / mu) is numeric data. Blast resistance (grade): Highly resistant, Resistant, Moderately resistant, Moderately susceptible, Susceptible, and Highly susceptible, is tabular data. When using tabular data, users typically select one of these as a candidate. Flowering date is date data. When evaluating date-related conditions, comparisons are typically made between dates before and after to determine quality. For example, if early flowering is preferred over late flowering, the candidate material's flowering date can be compared to see if it is earlier than the control material's, using the greater than, less than, or equal to criteria for comparison. Promotion or recommendation criteria typically include multiple numeric, tabular, and date data types, with promotion or recommendation primarily based on whether these conditions are met. Because a given condition only allows for one data type, determining which condition to compare also determines the data type, preventing comparisons between different data types.

[0115] It can be understood that different breeding materials with the same trait index should adopt the same weight, while different trait indexes can adopt different weights.

[0116] S104: Display analysis results, show promotion and recommendation details, eliminate breeding materials that do not meet the conditions, and make suggestions on the improvement direction of the materials.

[0117] (1) Visual display, showing promotion information.

[0118] like Figures 2 to 4 As shown: Users can see that in the model analysis process, from top to bottom, there are promotion analysis and recommendation analysis. In promotion analysis, each analysis node shows how many materials have passed the promotion, that is, they meet the promotion conditions. In recommendation analysis, each analysis node shows how many materials have been recommended, that is, they meet the recommendation conditions. For example: if the breeding material is rice, Figure 3 and Figure 4 The promotion analysis shown in is the promotion analysis part of the aforementioned promotion recommendation model. If the breeding material does not meet the promotion conditions, it will be eliminated. During the experiment, 10 materials were selected, namely material 1 to material 10. The trait indicators of the promotion conditions of this promotion analysis include rice blast resistance, panicle blast loss rate, fruit set rate and number of fruit set rate pilot tests. The test personnel input these measured data into the model analysis interface of the system. After input, the system will automatically calculate the specific values ​​of the rice blast resistance level, panicle blast loss rate, fruit set rate and number of fruit set rate pilot tests for each material according to the corresponding software algorithm, and display them in the interface, and also display the number of passes. If the material meets all the basic screening conditions, it will be determined as an promoted material. For example Figure 4 In the system, the materials that are promoted under the basic screening conditions include Material 1, Material 4, Material 5 and Material 9, while the remaining materials are eliminated because at least one of them does not meet the requirements. However, the eliminated materials are also given promotion coefficients and are distinguished by setting an "*" sign; because the system allows users to force the promotion of a certain material, the promotion coefficient is made for users' reference. If the user thinks that the material can be promoted, a forced passing operation can also be made to facilitate users to retain certain materials of special value.

[0119] The breeding materials that have been promoted will continue with the subsequent recommended analysis, such as Figure 4 As shown. Recommendation analysis can include multiple types of recommendations. This case includes recommended conditions for high and stable yields and disease-resistant materials. The recommended conditions for high and stable yields include the requirements of 5 trait indicators. The recommended coefficient is calculated according to the calculation method of the recommended coefficient in S103. Finally, the materials are ranked according to the recommended coefficient, that is, the materials with the highest recommended coefficient are ranked first and recommended to users first. Among them, those that meet the conditions compared with the model data are highlighted in one color, and those that do not meet the conditions are highlighted in another color. For example Figure 4 In the example, the "Rice Blast Loss Rate" value for Material 4 is 5%, which meets the recommended requirement of ≤15% and is highlighted in green. The "Average Yield per Mu - Control Percentage" value for Material 4 is 2%, which meets the recommended requirement of ≥10% and is highlighted in red. All other green-highlighted values ​​meet the recommended requirement and are therefore marked green, indicating that they meet the recommended requirement.

[0120] (2) Display recommendation details to users for materials that meet the recommendation conditions.

[0121] like Figure 5 As shown: The system summarizes the recommendation results for the user, calculates the recommendation coefficient for the user based on the weight of the model indicator, sorts the results according to the recommendation coefficient, recommends the best candidate materials to the user (such as breeders), and displays the recommendation details.

[0122] The system allows users to edit the weights of model indicators to better adapt to material selection and breeding. It is understandable that when the weights of promotion and recommendation conditions are changed, the promotion coefficient and recommendation coefficient will change accordingly. Based on different weight distributions, the same trait data can receive different promotion coefficients and recommendation coefficients, and thus the promotion ranking and recommendation ranking will change. This function can help users adjust the direction of promotion and recommendation based on the importance of phenotypic traits, helping users to better select distinctive breeding materials.

[0123] The system allows users to force a certain material to be upgraded. This function makes it easier for users to retain certain materials of special value.

[0124] according to Figure 5 The analysis results shown in the table show that "Material 1" and "Material 5" meet the recommendation conditions, with recommendation coefficients of 7.11 and 7.05 respectively. In order of recommendation, "Material 1" is recommended to the user first, followed by "Material 5". Click Figure 5 After confirmation in Figure 6 promotion and recommendation results.

[0125] Figure 6 The results of promotion and recommendation are displayed, and users are allowed to make final selections and confirmations. Users can make final confirmations on promotion and recommendation results based on the recommendation coefficient and specific trait values. Figure 7 , the system will give the user the final promotion and recommendation prompts.

[0126] After confirming the results, the system automatically pushes the promotion information and recommendation information to the database, which stores the relevant promotion information and recommendation information. The promotion and recommendation data enter the test set for model optimization.

[0127] (3) Provide suggestions for improving materials that do not meet the promotion requirements.

[0128] The system can also identify materials that don't meet certain criteria, identify the model conditions that don't meet them, and remind users to improve the corresponding properties. For example, if the "fruit set rate" of material 3 doesn't meet the requirements, the user will be reminded to improve it to become a better candidate material in the future.

[0129] Example 2

[0130] On the basis of Example 1, the following steps are also included:

[0131] S105: Optimize the existing promotion recommendation model.

[0132] Specific implementation reference Figure 9 、 Figure 10 ,exist Figure 9 In the interface shown, select the model to be optimized. After clicking Optimize, the system will analyze the promotion conditions and recommendation conditions in the model using the training and test datasets according to its internal analysis process. This includes:

[0133] (1) System based on Figure 8 Optimize the promotion recommendation model using the following process. For a detailed description of the process, refer to the instructions for "Optimizing the Promotion Recommendation Model Module."

[0134] (2) System operation as follows Figure 9 As shown, click a model that needs to be optimized to optimize the promotion recommendation model.

[0135] (3) The system automatically summarizes historical data and analyzes the trait indicators of the historical data of breeding materials.

[0136] (4) The analyzed data is used to form a training data set for analysis of promotion conditions and recommendation conditions. Figure 11 This is a screenshot of part of the training data set, which is used to analyze the promotion conditions and recommendation conditions.

[0137] (5) Calculate the threshold value of each trait through the analysis model of BioCloud. For the promotion conditions, the calculation method includes but is not limited to a single trait, and the threshold value needs to achieve a 5% material promotion pass rate. For the recommendation conditions, it includes but is not limited to recommending the top three materials among the promotion materials. For example, in this embodiment, the promotion condition is set to a 5% promotion pass rate, and the recommendation condition is set to allow the top three materials to be recommended.

[0138] (6) Optimize promotion conditions:

[0139] a) The calculated trait threshold is used as the candidate promotion condition for testing the test data set, such as Figure 12 As shown;

[0140] b) Manually set the pass rate of the test data set, for example, the promotion condition is set to a 95% pass rate;

[0141] c) When more than 95% of the materials in the test data set can be promoted, the candidate promotion conditions can be saved as optimized promotion conditions;

[0142] d) When the number of qualified materials in the test data set does not reach 95%, the threshold range will be readjusted and a new threshold will be recalculated based on BioCloud's analysis model. This calculation method includes, but is not limited to, readjusting the threshold of one or more properties based on the distribution of material data. After adjustment, continue testing the test data set until the appropriate optimization promotion conditions are found.

[0143] e) After the optimization results are determined, they are displayed in the system to help breeders understand the changes in conditions. Figure 10 The conditions before and after model optimization are shown in Figure 2. The average value of the promotion condition "Blast Resistance" remains unchanged, the highest level of "Blast Loss Rate" changes from ≤15% to ≤10%, the average value of "Seed Setting Rate" changes from ≥75% to ≥80%, and the number of pilot sites with "Seed Setting Rate" below 65% remains unchanged. Overall, the promotion conditions have become stricter. Although the promotion conditions have become stricter, this example still screens out breeding materials that meet the proportion (refer to Figure 13 The new promotion criteria will be used for future breeding selection, supporting breeders in selecting breeding materials using more stringent criteria.

[0144] (7) Optimize recommendation conditions:

[0145] a) The calculated trait threshold is used as a candidate recommendation condition for testing the test dataset, such as Figure 12 As shown;

[0146] b) Manually set the passing rate of the test data set. In this example, the recommendation condition is set to recommend the top three breeding materials, and only the top three materials are passed;

[0147] c) When the top three materials in the test dataset can be recommended, the candidate recommendation conditions can be saved as optimized recommendation conditions;

[0148] d) When the number of recommended materials in the test dataset does not meet the requirements, readjust the threshold range and recalculate new thresholds based on BioCloud's analysis model. Calculation methods include, but are not limited to, readjusting the thresholds of one or more properties based on the distribution of material data. After adjustment, continue testing the test dataset until the appropriate optimization recommendation conditions are optimized;

[0149] e) After the optimization results are determined, they are displayed in the system to help breeders understand the changes in conditions. Figure 10In the figure, the conditions before and after model optimization are shown. In this example, the average value of the recommended conditions for high and stable yield, "average yield per mu", is changed from ≥7% to ≥8%, the number of yield increase points of "average yield per mu" is changed from ≥7 to ≥10, the average value of "fruit setting rate" is changed from 75% to 80%, and the number of pilot sites with "fruit setting rate" below 65% remains unchanged. Overall, the recommended conditions have become more stringent. Although the recommended conditions have become more stringent, we still screened out breeding materials that meet the proportion (refer to Figure 14 The results of the screening data show that the overall performance of the materials in the system is constantly evolving, and breeding is effective. The new recommended conditions will be used for future breeding selection, and the system will recommend breeding materials to breeders using more stringent conditions.

[0150] (8) When both the promotion conditions and the recommendation conditions are optimized, the optimization of the promotion recommendation model is completed. Figure 10 As shown in the figure, after the model optimization is completed, the system will display the comparison results before and after optimization for the user's reference, and the user can save the optimized model data.

[0151] (9) If the optimized model is used for breeding analysis, the analysis process is consistent with the recommended model analysis process in the example embodiment.

[0152] (10) Integrate the indexed phenotypic data of the newly promoted materials into the new test dataset.

[0153] (11) Integrate the indexed phenotypic data of the newly added materials into the database as a training data set for the next analysis.

[0154] In other words, the system of the present invention supports the use of promotion recommendation models to analyze breeding materials for promotion recommendations, and also supports the optimization of promotion recommendation models using training and test data sets. Users can set promotion recommendation models based on their own needs and experience, and can also optimize existing promotion recommendation models. By utilizing the existing indexed phenotypic data in the system and utilizing big data and machine learning technologies to optimize the promotion recommendation model, the promotion conditions and recommendation conditions of breeding materials can be gradually improved, allowing the promotion recommendation model to continuously evolve with the development of breeding. This can better help breeders obtain higher-quality promotion recommendation models and screen better breeding materials.

[0155] In summary, the present invention completes the promotion and recommendation analysis of breeding materials according to the promotion recommendation model. On the one hand, it only needs to input test data to automatically perform promotion analysis and recommendation analysis on breeding materials, select promotion materials, and recommend excellent materials to users, saving time and effort. There is no need for breeders to rely on experience to check one by one, and there is no need to use Excel to organize data and write formulas, which greatly improves work efficiency. Thus, among a large number of trait indicators, the material with the best comprehensive index can be quickly screened out and recommended to breeders; on the other hand, for the new data imported, the system will automatically integrate and analyze the existing data and new data in the system, and the operation is very simple. In this way, various erroneous operations will not occur as in manual supplementation, thereby ensuring that as many high-quality materials as possible are retained, and accurate screening can be achieved through efficient data processing. Further, the system supports the optimization of the promotion recommendation model, helping breeders to obtain a better promotion recommendation model and screen better breeding materials more efficiently and accurately.

Claims

1. A method for automatically recommending crop breeding materials, characterized in that: The following steps are involved: S1: Obtain phenotypic data of breeding materials; S2: performing trait index analysis on the acquired phenotypic data to obtain trait index data; S3: Create a promotion recommendation model that allows users to customize promotion conditions and recommendation conditions. At the same time, the system combines the phenotypic indicator analysis data within the system into a training data set, uses the training data set to optimize the calculation of promotion conditions and recommendation conditions, and obtains an optimized promotion recommendation model through verification with a test data set; The method for obtaining the optimized promotion conditions includes: obtaining historical phenotypic data of breeding materials, analyzing the trait index of the historical data, and forming the analyzed data into a promotion training data set; Calculate the trait threshold of each trait indicator and use the trait threshold as the candidate promotion condition for testing the test data set; By manually setting the promotion pass ratio of the test data set, when the test data set exceeds this ratio, it will be promoted, and the candidate promotion conditions will be used as the optimized promotion conditions; if the number of promotion materials in the test data set does not reach this ratio, the corresponding trait threshold range will be readjusted, and the test data set will continue to be tested after the adjustment until the optimized promotion conditions are obtained; S4: Select the promotion recommendation model to process and analyze the breeding material trait index data. If the breeding material does not meet the promotion conditions, it will be eliminated. If the breeding material does not meet the promotion conditions and is eliminated, the user can perform forced promotion or forced recommendation operations. If the promotion conditions are met, the material will be promoted and obtain the promotion coefficient for ranking display. S5: Conduct recommendation analysis on the breeding materials that advance according to the recommendation conditions, calculate the recommendation coefficient, and recommend ranking the breeding materials that advance according to the recommendation coefficient; S6: Display analysis results.

2. The method for automatically recommending crop breeding materials according to claim 1, characterized in that: The promotion recommendation model includes promotion conditions and recommendation conditions; first, it is determined whether the various trait indicators of the breeding material meet the promotion conditions, and the promotion coefficient is calculated; if not, the breeding material is considered to be eliminated; If it is met, the analysis is carried out according to the recommended conditions and the recommendation coefficient is calculated; The score is calculated based on the number of promotion conditions and recommendation conditions met by the breeding materials. One point is added for each item that is met, and each item is assigned a corresponding weight to obtain the promotion coefficient or recommendation coefficient. Similarly, the scores of all promotion conditions or recommendation conditions are counted to summarize the promotion coefficient or recommendation coefficient. All breeding materials are ranked from high to low according to the recommendation coefficient. Based on the ranking, excellent materials are recommended to users.

3. The method for automatically recommending crop breeding materials according to claim 1, wherein: The optimized promotion recommendation model further includes optimized recommendation conditions, and the method for obtaining the optimized recommendation conditions includes: Obtain historical phenotypic data of breeding materials, perform trait index analysis on the historical data, organize the analyzed data into a recommended training data set, calculate the trait threshold of each trait index, and use the trait threshold as a candidate recommendation condition for testing the test data set; manually set the recommended ranking pass ratio of the test data set, and when the test data set exceeds this ratio, it will be promoted, and the candidate recommendation condition will be used as the optimized recommendation condition; if the number of recommended materials in the test data set does not reach this ratio, readjust the corresponding trait threshold range, and continue testing the test data set after adjustment until the optimized recommendation condition is obtained.

4. The method for automatically recommending crop breeding materials according to claim 2, wherein: The calculation method of the promotion coefficient or recommendation coefficient in the promotion recommendation model includes: converting the trait index data of a certain breeding material into a decimal greater than 0 and less than 1, then adding it to the "1" point representing that the conditions are met, and then multiplying it by the weight to use it as the recommendation coefficient; or first multiplying the converted decimal by the weight, and then adding "1" point to use it as the recommendation coefficient; when the value of a certain trait index is as small as possible, first convert the value into a decimal greater than 0 and less than 1, then subtract the decimal from 1, and then add it to the "1" point representing that the conditions are met, multiply it by the weight, or first multiply it by the weight, and then add "1" point to use it as the promotion coefficient or recommendation coefficient.

5. The method for automatically recommending crop breeding materials according to claim 1, 2 or 3, wherein: The trait indicator data include one or more of the following: mean value, control percentage, breeding value, breeding value ranking, minimum value, maximum value, least square mean, standard deviation, estimated value ranking, significance, F value, variance, number of yield increase points, and ratio of yield increase points.

6. A system for automatically recommending crop breeding materials according to the method for automatically recommending crop breeding materials according to any one of claims 1 to 5, characterized in that: include: A data acquisition module is used to obtain phenotypic data of breeding materials; A data analysis module is used to calculate the corresponding trait indicators of the phenotypic data; A data storage module, used to store phenotypic data and trait index data analyzed by the data analysis module; The promotion recommendation management module includes a promotion recommendation model management module and a promotion recommendation analysis task management module; the promotion recommendation model management module is used to create a promotion recommendation model; the promotion recommendation analysis task management module is used to create a promotion recommendation analysis task; The optimization and promotion recommendation model module includes a model optimization analysis module, a training data set, and a test data set; the model optimization analysis module is used for model optimization, including optimization of promotion conditions and recommendation conditions; the training data set is used to store training data integrated with the historical data of the model; The test data set is used to obtain promoted and recommended breeding materials, test the preliminary optimized model, and test whether the optimized model is suitable, and provide appropriate optimized promotion conditions and optimized recommendation conditions; The optimization method for promotion conditions includes: calculating the trait threshold of each trait indicator, using the trait threshold as a candidate promotion condition for testing the test data set; manually setting the promotion pass ratio of the test data set, and when the test data set exceeds the ratio, the candidate promotion condition is promoted, and the candidate promotion condition is used as the optimized promotion condition; if the number of promotion materials in the test data set does not reach the ratio, the corresponding trait threshold range is readjusted, and after the adjustment, the test data set is tested again until the optimized promotion condition is obtained; The rule engine processor is used to take the trait index data separated by the data analysis module as input, enter different rule models for operation, calculate the promotion coefficient and recommendation coefficient, and count the scores of all promotion conditions or recommendation conditions. It also ranks all breeding materials from high to low according to the recommendation coefficient; The result display module is used to display the analysis results.

7. The automatic promotion recommendation system for crop breeding materials according to claim 6, characterized in that: The rule engine processor includes a rule set area, a calculation area and a data area; The rule set area contains multiple rules that are set, and each rule contains specific business rule calculation logic; the rules refer to the trait indicator data range of the achievable promotion conditions and recommendation conditions set in the promotion recommendation model; The calculation area is used to calculate, summarize and rank the promotion coefficients and recommendation coefficients in the promotion recommendation model; The data area is used to store the data of each rule, mainly including rule parameters, calculation results and rule weights.

Images