An agricultural product recommendation system based on a data model

Abstract

The present application relates to the technical field of data processing, and more particularly to a kind of agricultural product recommendation system based on data model, the system includes: acquisition module, prediction module, event determination module, type determination module, grade determination module, adjustment module and output module.The present application comprehensively analyzes the dynamic data between crop irrigation frequency and growth height, combines the field quantity, keyword density and input speed and other behavior characteristics of user in input process, comprehensively reflects crop growth condition and user attention degree, and then accurately determines whether there is agricultural intervention demand, simultaneously, system fusion historical use frequency data, realizes the quantization evaluation of agricultural product applicability, effectively solves the problem of low accuracy and slow response speed when facing the dynamic change of target behavior due to too much dependence on static label and fixed sorting mechanism.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an agricultural product recommendation system based on a data model. Background Art

[0002] With the deepening of agricultural modernization and informatization, farmers are faced with temporal and spatial differences in multiple factors such as crop varieties, climatic conditions, soil types and growth stages during the production process. The irrigation, fertilization and pest control needs in each link are highly dynamic. At the same time, the agricultural inputs market has a wide variety of products and complex performance parameters. Although e-commerce platforms and social channels provide rich information, it is difficult to connect with field changes and individual needs in real time, resulting in information overload and delayed decision-making. Farmers often need to spend a lot of time comparing and judging when selecting and using agricultural inputs, making it difficult to achieve accurate and efficient production management.

[0003] Patent document with publication number CN108876562A discloses a product recommendation method and device for an agricultural supplies e-commerce platform. The method includes: obtaining at least one first associated tag group with corresponding season intervals and regional intervals based on the user's regional information and access time, the associated tag group including season intervals, regional intervals, use tags, category tags, and crop tags; obtaining at least one corresponding first information tag group based on the use tags, category tags, and crop tags in the first associated tag group, the information tag group including product names, category tags, use tags, and crop tags; filtering out a corresponding product information list from a product library based on the first information tag group; sorting the product information list according to a preset product sorting mechanism, obtaining a product recommendation sort, and feeding back to the user.

[0004] It can be seen that the product recommendation method and device of the agricultural supplies e-commerce platform have the following problems: the patent only relies on five static labels of "region + season + purpose + category + crop" for screening, which cannot reflect the multi-dimensional changes in the agricultural production process; the product list is only sorted once through a preset sorting mechanism (such as by label matching or fixed weight), and the recommendation logic cannot be dynamically adjusted according to the frequency of use and effect feedback. Summary of the Invention

[0005] To this end, the present invention provides an agricultural product recommendation system based on a data model, which is used to overcome the problems of low accuracy and slow response speed in the existing technology when facing dynamic changes in the target's behavior due to over-reliance on static labels and fixed sorting mechanisms by real-time collection of multi-dimensional dynamic data and dynamic adjustment of thresholds.

[0006] To achieve the above objectives, the present invention provides an agricultural product recommendation system based on a data model, comprising:

[0007] an acquisition module, configured to acquire, in real time, all crop irrigation frequencies, all crop growth heights, a field quantity when input description content is input, a keyword density, an input speed, and a historical use frequency of each agricultural product in a preset product library within a preset recording time length when a target starts to be input;

[0008] a prediction module, connected with the acquisition module, configured to predict a severity threshold value according to a preset agricultural model, all the crop irrigation frequencies, and all the crop growth heights;

[0009] an event determination module, connected with the acquisition module and the prediction module respectively, configured to determine an agricultural intervention demand event according to the field quantity, all the crop irrigation frequencies, and the severity threshold value, to form an intervention determination result;

[0010] a type determination module, connected with the acquisition module and the event determination module respectively, configured to determine a recommended interaction type of a current target according to the intervention determination result and the input speed;

[0011] a grade determination module, connected with the acquisition module and the type determination module respectively, configured to determine a priority grade of this time recommendation according to the recommended interaction type, the crop growth height, the keyword density, and the field quantity;

[0012] an adjustment module, connected with the grade determination module and the prediction module respectively, configured to adjust the severity threshold value according to a number of times that the priority grade appears within a preset adjustment time length, to form an adjusted threshold value;

[0013] an output module, connected with the type determination module, configured to output a recommendation result based on the recommended interaction type determined again through the adjusted threshold value and the historical use frequency.

[0014] Further, the event determination module comprises:

[0015] a mean value calculation unit, configured to calculate a mean value of all the crop irrigation frequencies, to form a mean irrigation frequency;

[0016] a first normalization unit, connected with the mean value calculation unit, configured to perform normalization processing on the mean irrigation frequency, to obtain a normalized irrigation frequency, and to perform normalization processing on the field quantity, to obtain a normalized field quantity;

[0017] an intervention index calculation unit, connected with the first normalization unit, configured to perform weighted summation calculation on the normalized irrigation frequency, a preset irrigation frequency weight, the normalized field quantity, and a preset field quantity weight, to obtain an intervention index;

[0018] An event determination unit is connected to the intervention index calculation unit and is used to determine the occurrence of the agricultural input intervention demand event based on the intervention index and the severity threshold, and form the intervention determination result.

[0019] Furthermore, the event determination unit includes:

[0020] A first determination subunit is configured to determine that the agricultural input intervention demand event occurs when the intervention index is greater than the severity threshold, and form the intervention determination result;

[0021] a frequency change calculation subunit, connected to the first determination subunit, for calculating, when the intervention index is less than or equal to the severity threshold, the absolute value of the difference between the crop irrigation frequencies in all adjacent preset time windows within the preset recording duration to form a plurality of frequency change rates;

[0022] a change fluctuation calculation subunit, connected to the frequency change calculation subunit, for calculating the standard deviation of all the frequency change rates to form a change rate fluctuation value;

[0023] The second determination subunit is connected to the change fluctuation calculation subunit and is used to determine that the agricultural input intervention demand event occurs when the change rate fluctuation value is greater than a preset change rate fluctuation threshold, and form the intervention determination result.

[0024] Furthermore, the type determination module includes:

[0025] a speed comparison unit, configured to compare the input speed with a preset speed threshold when forming the intervention determination result, to form a speed comparison result;

[0026] A type determination unit is connected to the speed comparison unit and is configured to determine that the recommended interaction type is an emergency intervention type when the speed comparison result shows that the input speed is greater than the preset speed threshold.

[0027] Furthermore, the level determination module includes:

[0028] a strength determination unit, configured to determine the strength of the recommendation intention according to the keyword density and the number of fields when the emergency intervention type is determined;

[0029] A priority level determination unit is connected to the strength determination unit and is used to determine the response level of the current recommendation as the priority level according to the strength of the recommendation intention and the growth height of the crop.

[0030] Furthermore, the strength determination unit includes:

[0031] a density fluctuation calculation subunit, configured to calculate the standard deviation of the density of all the keywords within the preset recording time to obtain a density fluctuation value;

[0032] A word count fluctuation calculation subunit, configured to calculate a standard deviation of the number of all fields within the preset recording time to obtain a word count fluctuation value;

[0033] a normalization subunit, connected to the density fluctuation calculation subunit and the word count fluctuation calculation subunit, respectively, for normalizing the density fluctuation value to obtain a density normalized fluctuation value, and normalizing the word count fluctuation value to obtain a word count normalized fluctuation value;

[0034] a consistency calculation subunit, connected to the normalization subunit, for calculating a correlation coefficient between the density normalized fluctuation value and the word count normalized fluctuation value to obtain a fluctuation consistency;

[0035] The strength determination subunit is connected to the consistency calculation subunit and is used to calculate the relative deviation between the fluctuation consistency and the preset standard consistency when the fluctuation consistency is greater than the preset standard consistency, so as to determine the strength of the recommendation intention.

[0036] Furthermore, the priority determination unit includes:

[0037] a height change calculation subunit, configured to calculate, when the strength of the recommendation intention is greater than a preset strength threshold, the absolute value of the difference between the crop growth heights in all adjacent preset time windows to obtain a plurality of height change rates;

[0038] an altitude change fluctuation calculation subunit, connected to the altitude change calculation subunit, for calculating the standard deviation of all the altitude change rates to form an altitude change fluctuation value;

[0039] The priority level determination subunit is connected to the altitude change fluctuation calculation subunit and is used to determine that the response level is a priority level when the altitude change fluctuation value is greater than a preset altitude change fluctuation threshold.

[0040] Furthermore, the adjustment module includes:

[0041] a recording unit, configured to record a current timestamp to form an initial time point when the priority level occurs more than a preset threshold number of times;

[0042] a number acquisition unit connected to the recording unit, configured to acquire the number of times within a time period starting from the initial time point to each time stamp within the preset adjustment time period, to form a plurality of times;

[0043] a frequency fluctuation calculation unit, connected to the frequency acquisition unit, for calculating the standard deviation of all the frequencies to form a frequency fluctuation value;

[0044] An adjusting unit is used to adjust the severity threshold according to the number fluctuation value to form an adjusted threshold.

[0045] Furthermore, the adjustment unit includes:

[0046] A number fluctuation comparison subunit is used to compare the number fluctuation value with a preset number fluctuation threshold to form a number fluctuation comparison result;

[0047] An adjustment subunit is connected to the number comparison subunit and is used to increase the severity threshold according to the relative deviation between the number fluctuation value and the preset number fluctuation threshold and the preset adjustment coefficient to form an adjustment threshold when the number fluctuation comparison result is that the number fluctuation value is greater than the preset number fluctuation threshold.

[0048] Furthermore, the output module includes:

[0049] a retrieval unit, configured to retrieve a plurality of agricultural products from the preset product library according to the re-determined recommendation interaction type, so as to obtain a plurality of candidate products;

[0050] a usage change rate calculation unit connected to the retrieval unit, for calculating the difference in the historical usage frequency of each candidate product in each adjacent preset time window within a preset historical duration to obtain a plurality of usage change rates;

[0051] a usage change fluctuation calculation unit connected to the usage change rate calculation unit, for calculating a standard deviation of all the usage change rates to obtain a usage change fluctuation value;

[0052] a sorting unit connected to the usage change fluctuation calculation unit, for sorting the usage change fluctuation values ​​of all the candidate products from small to large to form a product list;

[0053] An output unit is connected to the ranking unit and is used to recommend the candidate products with the first preset ranking number in the product list to output the recommendation result.

[0054] Compared with the prior art, the present application has the beneficial effects that, by comprehensively analyzing the dynamic data between the irrigation frequency of crops and the growth height, combining the behavior characteristics of the user in the input process such as the number of fields, the keyword density and the input speed, the growth condition of crops and the user attention degree are comprehensively reflected, and then it is accurately determined whether there is a demand for agricultural intervention, at the same time, the system integrates the historical use frequency data to realize the quantitative evaluation of the applicability of agricultural products, on this basis, through the dynamic adjustment of the recommended interaction type and the priority level, it is ensured that the system can output more targeted and timely recommendation results under different emergency degrees and demand backgrounds, thereby improving the efficiency and effect of agricultural use, reducing resource waste, and effectively solving the problems of low accuracy and slow response speed in recommendation when facing the dynamic changes of target behavior due to too much dependence on static labels and fixed sorting mechanism.

[0055] Further, by normalizing and weighting the average irrigation frequency and the user description behavior (such as the number of fields), a comprehensive intervention index considering the actual growth of crops and the subjective input behavior of the user is constructed, and then the precise intervention demand judgment mechanism is formed in combination with the severity threshold. This way effectively reflects the internal correlation between crop water demand changes and user attention, making the intervention judgment more close to the actual agricultural conditions, which is beneficial to the system to identify potential problems and output targeted recommendation results in time.

[0056] Further, by the double judgment mechanism, the dynamic fluctuation of the intervention index and the irrigation frequency change is fully utilized, and the comprehensive judgment is made through the static intensity index (intervention index) and the dynamic change trend (frequency fluctuation), which can significantly improve the sensitivity and accuracy of abnormal water use behavior, avoid misjudgment due to short-term fluctuations or neglect due to long-term stability but abnormal trend, make the organic linkage between various data parameters, and form a more responsive and robust agricultural recommendation logic.

[0057] Further, by comparing the input speed with the preset threshold, the system can dynamically identify the urgency of the user input behavior, so as to match the recommended interaction type with the actual intervention scene, improve the timeliness and fitting degree of the recommended response. This mechanism uses "input behavior" as an implicit characteristic variable to indirectly map the potential emergency demand of the user, so that the recommendation process not only depends on objective data (such as irrigation frequency), but also can perceive the subjective urgency in the user's expression, effectively enhancing the intelligent judgment and adaptability of the agricultural recommendation system.

[0058] Furthermore, through the logical association between input features (such as keyword density and number of fields) and crop status data (growth height), a dual evaluation mechanism of the "urgency" of input content and the "current demand" of crops was established, ensuring that the recommendation system not only responds quickly, but also recommends content that is accurate and in line with the scenario, thereby effectively improving the timeliness and scientific nature of agricultural input recommendations and avoiding waste of recommendation resources or intervention delays.

[0059] Furthermore, by calculating the fluctuations in keyword density and field count, the system can identify the consistency and urgency of user input. When density and word count fluctuate with high consistency, it indicates a more specific and urgent user need, thus increasing the strength of the recommendation intent response. By quantifying input variations, the system accurately captures the strength of the user's recommendation needs, ensuring that the system prioritizes more urgent and specific needs, thereby improving the efficiency and accuracy of the recommendation system.

[0060] Furthermore, by analyzing changes in crop height, the system can determine crop growth status, ensuring that high-need agricultural interventions are prioritized in situations of significant crop growth fluctuations. By setting preset intensity thresholds and height fluctuation thresholds, the system intelligently identifies situations requiring immediate intervention, avoiding wasteful resources and improving the efficiency of agricultural intervention responses. Furthermore, the logical correlation between data, such as the link between crop growth fluctuations and intervention needs, further enhances the accuracy and responsiveness of the recommendation system through analysis of height change rate and volatility.

[0061] Furthermore, through the fluctuations in the frequency of occurrence of directional priority levels, the system can dynamically adjust the severity threshold. If the priority level appears frequently and fluctuates greatly within a certain period of time, it means that the need for intervention may increase. The system will strengthen the response by adjusting the threshold, ensuring that the system can flexibly respond to various changing intervention needs, thereby achieving more accurate agricultural intervention, optimizing crop growth management, and effectively avoiding waste of resources or excessive intervention.

[0062] Furthermore, by comparing the frequency fluctuation value with the preset frequency fluctuation threshold and adjusting the parameters based on the relative deviation and preset adjustment coefficient, the severity threshold can automatically respond to changes based on the fluctuation of actual intervention needs. This mechanism enhances the system's ability to perceive changes in abnormal frequency and achieves adaptive optimization of intervention sensitivity, ensuring that the response threshold is promptly raised when abnormalities fluctuate frequently, reducing misjudgments and intervention redundancy, and improving the scientific nature and stability of intervention strategies.

[0063] Further, by correlating multiple data such as recommended interaction type, historical usage frequency, change rate and fluctuation, the recommended result is not only based on demand type matching, but also considers the stability and trend of historical use of agricultural products, thereby improving the scientificity, adaptability and credibility of recommended products, and helping users to more reasonably select agricultural products in different situations. BRIEF DESCRIPTION OF DRAWINGS

[0064] Figure 1 a schematic diagram of the agricultural product recommendation system based on a data model of the embodiment;

[0065] Figure 2 a determination logic diagram for determining that an agricultural intervention demand event occurs in the event determination unit of the embodiment;

[0066] Figure 3 a determination logic diagram for determining the recommended interaction type in the type determination unit of the embodiment;

[0067] Figure 4 a determination logic diagram for determining the recommended intention strength in the strength determination subunit of the embodiment. DETAILED DESCRIPTION

[0068] In order to make the objects and advantages of the present application clearer, the present application will be further described below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0069] The preferred embodiments of the present application will be described below with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are only used to explain the technical principles of the present application and do not limit the protection scope of the present application.

[0070] Please refer to Figure 1 shown, which is a schematic diagram of the agricultural product recommendation system based on a data model of the embodiment;

[0071] In the national agricultural service cloud platform, by connecting various agricultural service terminals, long-term and continuous collection and accumulation of farmers' procurement records, use feedback and effect evaluation of various agricultural products (such as seeds, fertilizers, pesticides, etc.), a widely covered and real-time updated historical usage frequency database is constructed. Even in the case of new products just online, there is no enough use data, intelligent completion can be made according to the product category, applicable crops and the use frequency average of similar historical products, to ensure that all products have basic comparability, facilitate the stable operation of the recommendation engine, so as to realize accurate recommendation across regions and crops, and support efficient landing of agricultural intelligent decision and intervention response.

[0072] Based on the national agricultural service cloud platform, the present embodiment provides an agricultural product recommendation system based on a data model, comprising:

[0073] The acquisition module is used to obtain in real time the irrigation frequency of all crops, the growth height of all crops, the number of fields when entering description content, keyword density, input speed, and the historical usage frequency of each agricultural product in the preset product library within a preset recording period from the start of target input;

[0074] a prediction module connected to the acquisition module, configured to predict a severity threshold based on a preset agricultural input model, the irrigation frequency of all the crops, and the growth height of all the crops;

[0075] an event determination module, connected to the acquisition module and the prediction module respectively, for determining the occurrence of an agricultural input intervention demand event based on the number of fields, the irrigation frequency of all crops, and the severity threshold, and forming an intervention determination result;

[0076] a type determination module, connected to the acquisition module and the event determination module respectively, for determining a recommended interaction type for the current target based on the intervention determination result and the input speed;

[0077] a level determination module, connected to the acquisition module and the type determination module respectively, for determining the priority level of this recommendation according to the recommended interaction type, the crop growth height, the keyword density, and the number of fields;

[0078] an adjustment module, connected to the level determination module and the prediction module respectively, for adjusting the severity threshold according to the number of occurrences of the priority level within a preset adjustment time period to form an adjustment threshold;

[0079] An output module is connected to the type determination module and is used to output a recommendation result based on the recommended interaction type re-determined by the adjustment threshold and the historical usage frequency.

[0080] "Targets" are users who currently use the agricultural input service cloud platform to input demand information. The system collects their input data and related environmental variables to understand their demand characteristics and generate personalized recommendations.

[0081] Crop irrigation frequency refers to the average number of times a user irrigates each crop per day, reflecting the user's emphasis on crop water management and the crop's likely growth stage. A higher frequency indicates that the crop is in a critical growth period.

[0082] Crop growth height refers to the average daily recorded height of all crops planted by the user. It is used to characterize crop growth status and monitor its health and development stage.

[0083] The description content refers to the textual description of agricultural input needs that users actively fill out on the platform (including but not limited to "tomato leaves turning yellow" and "need to control weeds"), which is an important source of information for users to subjectively express their needs.

[0084] The number of fields refers to the number of independent fields (or data items) used by users during the input process, including but not limited to “crop type”, “crop area”, and “problem description”, reflecting the degree of structure of the user input content.

[0085] Keyword density refers to the proportion of agriculture-related keywords (including but not limited to "leaf spots", "pests and diseases", and "yellowing") in the text input by the user. It is used to evaluate the professionalism and clarity of the input content and helps improve the accuracy of recommendations.

[0086] Input speed refers to the average time rate at which a user inputs the entire description content. It reflects the user's urgency and indirectly assesses their information preparation level.

[0087] Historical usage frequency refers to the historical frequency of various agricultural inputs (including but not limited to a certain brand of fertilizers and herbicides) used by users as recorded in the National Agricultural Inputs Service Cloud Platform. The statistical dimension can be national or regional, which can reflect the popularity and scope of application of the product and provide reference priority for the recommendation system.

[0088] The preset product library refers to a collection of agricultural product information pre-built and continuously updated by the system, including but not limited to basic information on various agricultural products, applicable crop types, usage frequency, and historical feedback records. Its configuration depends on factors such as the target crop type, regional agricultural practices, user preferences, and past usage data, and typically includes dozens to hundreds of products. In this embodiment, the preset product library covers approximately 200 agricultural products, which enables the system to quickly retrieve highly matching candidate products after generating a recommended interaction type, providing basic support for subsequent sorting and recommendation output, significantly improving the pertinence and response efficiency of recommendations.

[0089] The acquisition module collects data through the real-time operation behavior of users on the national agricultural input service cloud platform: when the user logs in and fills in the description of agricultural input needs on the page, the system automatically records the daily irrigation frequency and growth height of all crops planted by the user within the preset recording time (provided by the agricultural equipment bound to the user or the data uploaded by the platform), and at the same time counts the number of fields, keyword density and input speed in the process of inputting the description (obtained through real-time analysis of keyboard events and text processing); in addition, the system also synchronously extracts the historical usage frequency of various agricultural input products from the built-in product library for subsequent recommendation analysis and matching judgment.

[0090] The default agricultural input model is the gradient boosting decision tree model, and its model architecture is:

[0091] Architecture composition:

[0092] Input features:

[0093] Crop irrigation frequency (daily average), crop growth height (daily average growth rate, current height), crop species labels (can be one-hot encoded or embedded), and historical recommendation feedback data (to improve model robustness);

[0094] Weak learner: A regression tree consisting of dozens of CART (Classification and Regression Tree), each tree learns the residual of the previous step prediction.

[0095] Loss function: Minimize the mean square error (MSE) as the regression target.

[0096] Iterative strategy: Continuously approach the actual severity value through each round of residual fitting.

[0097] Model training process:

[0098] Data preparation:

[0099] Historical farm data: sensor records from user-bound devices (e.g., irrigation records, crop height);

[0100] Label generation: The "intervention severity" value (numeric label) is provided by agricultural experts or marked through existing agricultural input use feedback systems;

[0101] Data cleaning and feature engineering:

[0102] Filling missing values ​​(e.g., mean, median); normalizing and standardizing features (e.g., standardizing growth height); unfolding time series into static features (e.g., mean / fluctuation of irrigation frequency over the past three days);

[0103] Model training parameter settings:

[0104] num_leaves=64: controls the complexity of each tree;

[0105] learning_rate=0.05: lower learning rate to improve robustness;

[0106] n_estimators=1000: maximum number of trees;

[0107] early_stopping_rounds=50: prevent overfitting;

[0108] Training strategy:

[0109] Use five-fold cross validation to ensure generalization ability;

[0110] Parameter adjustment method: based on grid search or Bayesian optimization;

[0111] Output:

[0112] The model will output a floating-point value "severity prediction value" for subsequent intervention decisions.

[0113] An agricultural input intervention demand event occurs when the system determines that the current crop growth situation presents risks requiring agricultural input intervention, including but not limited to diseases, high temperature and water shortage, and nutritional imbalance.

[0114] The recommended interaction type is determined by the system automatically determining the presentation method and content type of recommendations based on user input and the type of intervention event. In this embodiment, the emergency intervention type is determined by the system immediately proactively recommending a highly effective and targeted agricultural input product portfolio based on the predicted results after determining that an intervention event has occurred, prioritizing the user's need for rapid response in crisis scenarios.

[0115] Priority level means that in emergency intervention interaction scenarios, the system will automatically assign the highest priority to the recommendation process to ensure that it runs at the highest response level in the recommendation engine.

[0116] The preset recording duration refers to the time window used by the system to collect data such as crop irrigation frequency and crop growth height. It depends on the crop growth cycle, irrigation frequency and data collection requirements. It is usually set between 6 and 30 days. In this embodiment, it is set to 7 days, which can provide sufficient data to analyze the irrigation needs and growth trends of crops in a shorter period.

[0117] The preset adjustment duration is the time window during which the system adjusts the severity threshold when the intervention index reaches a preset threshold. This window depends on the crop growth cycle and intervention frequency, ensuring that threshold adjustments can promptly reflect changing needs. It is typically set between 30 seconds and 2 minutes, but in this example, it is set to 1 minute. This allows for rapid adjustment of the severity threshold based on frequent intervention needs in a short period of time, ensuring a more flexible and agile system response to crop growth conditions.

[0118] The acquisition module collects the crop irrigation frequency, growth height of the target user within a preset time length, and input behavior related parameters and historical product usage frequency data in real time, the prediction module calculates the severity threshold of the current crop problem based on the agricultural model, the event determination module combines the field quantity and threshold result to determine whether to trigger the agricultural intervention event, the type determination module determines the emergency level of the recommended interaction according to the input speed and intervention situation, the level determination module further evaluates the recommended priority according to the crop state and user input characteristics, and on this basis, the adjustment module dynamically corrects the severity determination standard combined with the historical response frequency, and finally the output module generates the agricultural product recommendation result which is accurate and matches the emergency level based on the recommended type and historical usage frequency.

[0119] By comprehensively analyzing the dynamic data between crop irrigation frequency and growth height, combining the field quantity, keyword density and input speed and other behavior characteristics of the user in the input process, the crop growth condition and the user attention degree are comprehensively reflected, and then it is accurately determined whether there is an agricultural intervention demand, at the same time, the system integrates the historical usage frequency data to realize the quantitative evaluation of the applicability of agricultural products, and on this basis, through the dynamic adjustment of the recommended interaction type and priority level, it ensures that the system can output more targeted and timely recommendation results under different emergency levels and demand backgrounds, thereby improving the efficiency and effect of agricultural use, reducing resource waste, and effectively solving the problems of low accuracy and slow response speed in recommendation when facing the dynamic changes of the target due to excessive dependence on static tags and fixed sorting mechanism.

[0120] Specifically, the event determination module comprises:

[0121] The mean value calculation unit is used to calculate the average value of all the crop irrigation frequencies to form the average irrigation frequency.

[0122] The first normalization unit is connected with the mean value calculation unit, used to normalize the average irrigation frequency to obtain the normalized irrigation frequency, and normalize the field quantity to obtain the field normalized quantity.

[0123] The intervention index calculation unit is connected with the first normalization unit, used to calculate the weighted sum of the normalized irrigation frequency, the preset irrigation frequency weight, the field normalized quantity and the preset field quantity weight to obtain the intervention index.

[0124] The event determination unit is connected with the intervention index calculation unit, used to determine whether the agricultural intervention demand event occurs according to the intervention index and the severity threshold to form the intervention determination result.

[0125] The preset irrigation frequency weight refers to the importance of crop irrigation frequency in the overall judgment when calculating the intervention index, which depends on the correlation between irrigation frequency and crop growth status and the demand for agricultural material use. It is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.6, which can enhance the sensitivity of the system to the water management of crops, facilitating the rapid identification of agricultural material intervention needs caused by irrigation abnormalities.

[0126] The preset field quantity weight refers to the weight of the number of fields in the input description on the overall judgment. It reflects the influence of the user's description on the judgment of intervention needs. The value is usually set between 0.3 and 0.7. In this embodiment, it is set to 0.4, which can balance the contribution of the completeness of user input to the accuracy of the recommendation and improve the judgment ability of the system in the case of limited input.

[0127] First, the mean value calculation unit calculates the average irrigation frequency of the crop within the preset recording time, forming the average irrigation frequency. Then, the first normalization unit normalizes the average irrigation frequency and the number of fields input by the user to obtain comparable standardized data. Next, the intervention index calculation unit combines the normalized irrigation frequency and the number of fields, and applies the preset weights respectively, and performs weighted summation to obtain the intervention index reflecting the emergency degree of the current agricultural condition. Finally, the event determination unit compares the index with the severity threshold provided by the prediction module to determine whether there is an agricultural material intervention demand event, and generates the corresponding intervention determination result.

[0128] By normalizing and weighting the average irrigation frequency and the user's description behavior (such as the number of fields), a comprehensive intervention index is constructed that takes into account the actual growth of crops and the subjective input behavior of users, and then combined with the severity threshold to form a precise intervention demand determination mechanism. This way effectively reflects the internal relationship between crop water demand changes and user attention, making the intervention judgment more in line with the actual agricultural condition, which is conducive to the system to identify potential problems and output targeted recommendation results in a timely manner.

[0129] Please continue to refer to Figure 2 shown, which is the determination logic diagram of the event determination unit of the present embodiment for determining whether an agricultural material intervention demand event occurs;

[0130] The event determination unit comprises:

[0131] The first determination sub-unit is configured to determine that the agricultural material intervention demand event occurs when the intervention index is greater than the severity threshold, and form the intervention determination result;

[0132] a frequency change calculation subunit, connected to the first determination subunit, for calculating, when the intervention index is less than or equal to the severity threshold, the absolute value of the difference between the crop irrigation frequencies in all adjacent preset time windows within the preset recording duration to form a plurality of frequency change rates;

[0133] a change fluctuation calculation subunit, connected to the frequency change calculation subunit, for calculating the standard deviation of all the frequency change rates to form a change rate fluctuation value;

[0134] The second determination subunit is connected to the change fluctuation calculation subunit and is used to determine that the agricultural input intervention demand event occurs when the change rate fluctuation value is greater than a preset change rate fluctuation threshold, and form the intervention determination result.

[0135] The preset time window refers to the time granularity unit used to calculate changes in crop irrigation frequency. It means that within the entire preset recording duration of the target input, the time is divided into several time periods (windows) of equal length, so as to quantitatively analyze the data changes between adjacent time periods. It depends on the crop type, growth cycle and data collection frequency. In this embodiment, it is set to 1 day, that is, the system divides the preset recording duration of the target input into several time periods in units of "days" for analyzing changes in daily irrigation frequency.

[0136] First, the first determination subunit determines whether the intervention index exceeds the severity threshold. If so, it is directly identified as an intervention-required event. If not, the frequency change calculation subunit analyzes the changes in crop irrigation frequency between all adjacent time windows within the target input duration and calculates the frequency change rate for each adjacent time period. Next, the change fluctuation calculation subunit further calculates the standard deviation of these rates to form a change rate fluctuation value. Finally, the second determination subunit determines whether this fluctuation value exceeds a preset threshold. If so, it is still identified as an intervention-required event.

[0137] The dual judgment mechanism fully utilizes the dynamic volatility of the intervention index and irrigation frequency changes. Comprehensive judgment is made through the two dimensions of static intensity index (intervention index) and dynamic change trend (frequency fluctuation). It can significantly improve the sensitivity and judgment accuracy of abnormal water use behavior, avoid misjudgment due to short-term fluctuations or ignore abnormal trends due to long-term stability, and form an organic linkage between the various data parameters to form a more responsive and robust agricultural input recommendation logic.

[0138] Please continue reading Figure 3 As shown, it is a decision logic diagram of the type determination unit in this embodiment for determining the recommended interaction type;

[0139] The type determination module includes:

[0140] a speed comparison unit configured to compare the input speed with a preset speed threshold to form a speed comparison result when forming the intervention determination result;

[0141] a type determination unit connected with the speed comparison unit and configured to determine that the recommended interaction type is an emergency intervention type when the speed comparison result is that the input speed is greater than the preset speed threshold.

[0142] The preset speed threshold is a reference benchmark used by the system to measure whether the user input behavior is in an "emergency state", which depends on the average input speed distribution of past users of the platform in different use scenarios, and is set in combination with the sensitivity of the actual demand response scenario, and is usually set between 3 characters / second and 8 characters / second, and is set to 5 characters / second in the embodiment, so as to realize fast response and priority processing, improve user experience and the practical value of the recommendation system.

[0143] After the intervention determination result is generated, the speed comparison unit first compares the speed of the target user when inputting the description content with the preset speed threshold set by the system to form a speed comparison result. If the input speed is higher than the preset threshold, the type determination unit determines that the current recommended task is an emergency intervention type, so that the product pushing mode with fast, clear and strong responsiveness is preferentially used in the recommendation strategy to meet the urgent needs of the user for agricultural intervention.

[0144] Through the comparison between the input speed and the preset threshold, the system can dynamically identify the urgency of the user input behavior, so as to match the recommended interaction type with the actual intervention scene and improve the timeliness and fitting degree of the recommended response. This mechanism uses "input behavior" as an implicit characteristic variable to indirectly map the potential emergency demand of the user, so that the recommendation process not only depends on objective data (such as irrigation frequency), but also can perceive the subjective urgency in the user's expression, effectively enhancing the intelligent judgment and adaptability of the agricultural recommendation system.

[0145] Specifically, the level determination module includes:

[0146] a strength determination unit configured to determine a recommendation intention strength according to the keyword density and the field quantity when the emergency intervention type is determined;

[0147] a priority level determination unit connected with the strength determination unit and configured to determine that the response level of the current recommendation is a priority level according to the recommendation intention strength and the crop growth height.

[0148] The recommended intention intensity is an index for measuring the urgency and explicitness of the user input, and the level determination module confirms the recommended response level through a two-level determination process. First, the intensity determination unit determines the recommended intention intensity of the user expression according to the keyword density and the field quantity in the input content after the user input is identified as an “urgent intervention type”. The higher the keyword density and the more the field quantity, the more specific and explicit the user expression. Then, the priority level determination unit determines the priority level of the current recommendation by combining the recommended intention intensity and the growth height of the current crop (reflecting the stage of the crop and the urgency of the agricultural material demand), so as to determine whether the system should push the result and provide intervention suggestions in priority.

[0149] Through the logical association between the input features (such as keyword density and field quantity) and the crop state data (growth height), a dual evaluation mechanism of “urgency” of input content and “current demand” of crop is established, which ensures that the recommendation system not only responds quickly, but also provides accurate and scenario-compliant recommendation content, thereby effectively improving the timeliness and scientificity of agricultural material recommendation and avoiding waste of recommended resources or delay of intervention.

[0150] Please continue to refer to Figure 4 As shown in FIG. 6, which is a determination logic diagram for determining the recommended intention intensity by the intensity determination sub-unit in the embodiment;

[0151] The intensity determination unit comprises:

[0152] The density fluctuation calculation sub-unit is configured to calculate the standard deviation of all the keyword densities within the preset recording time length to obtain a density fluctuation value.

[0153] The word number fluctuation calculation sub-unit is configured to calculate the standard deviation of all the field quantities within the preset recording time length to obtain a word number fluctuation value.

[0154] The normalization sub-unit is connected with the density fluctuation calculation sub-unit and the word number fluctuation calculation sub-unit, and is configured to normalize the density fluctuation value to obtain a normalized density fluctuation value, and normalize the word number fluctuation value to obtain a normalized word number fluctuation value.

[0155] The consistency calculation sub-unit is connected with the normalization sub-unit, and is configured to calculate the correlation coefficient of the normalized density fluctuation value and the normalized word number fluctuation value to obtain a fluctuation consistency degree.

[0156] The intensity determination sub-unit is connected with the consistency calculation sub-unit, and is configured to calculate the relative deviation between the fluctuation consistency degree and the preset standard consistency degree to determine the recommended intention intensity when the fluctuation consistency degree is greater than the preset standard consistency degree.

[0157] The preset standard consistency is a threshold used to measure the consistency of user input content. It depends on the characteristics of the input content, the system's requirements for consistency, and historical data feedback. It is usually set between 0.7 and 0.9. In this embodiment, it is set to 0.85, which can ensure that the system responds more sensitively and efficiently to input with strong consistency, thereby optimizing the recommendation effect and improving the user experience.

[0158] First, the density fluctuation calculation subunit calculates the standard deviation of the density of all keywords in the user input to form a density fluctuation value. Next, the word count fluctuation calculation subunit calculates the standard deviation of the number of all fields in the input content to obtain a word count fluctuation value. The normalization subunit then normalizes the density fluctuation value and the word count fluctuation value to obtain a normalized density fluctuation value and a normalized word count fluctuation value. Subsequently, the consistency calculation subunit calculates the correlation coefficient between the two to form a fluctuation consistency. Finally, the strength determination subunit determines the strength of the recommendation intent based on the relative deviation between the fluctuation consistency and the preset standard consistency.

[0159] By calculating the fluctuations in keyword density and field count, the system can identify the consistency and urgency of user input. When density and word count fluctuate with high consistency, it indicates a more specific and urgent user need, thus increasing the strength of the recommendation intent response. By quantifying input variations, the system accurately captures the strength of the user's recommendation needs, ensuring that the system prioritizes more urgent and specific needs, thereby improving the efficiency and accuracy of the recommendation system.

[0160] Specifically, the priority determination unit includes:

[0161] a height change calculation subunit, configured to calculate, when the strength of the recommendation intention is greater than a preset strength threshold, the absolute value of the difference between the crop growth heights in all adjacent preset time windows to obtain a plurality of height change rates;

[0162] an altitude change fluctuation calculation subunit, connected to the altitude change calculation subunit, for calculating the standard deviation of all the altitude change rates to form an altitude change fluctuation value;

[0163] The priority level determination subunit is connected to the altitude change fluctuation calculation subunit and is used to determine that the response level is a priority level when the altitude change fluctuation value is greater than a preset altitude change fluctuation threshold.

[0164] The preset intensity threshold refers to a critical value set by the system when judging the intensity of the recommendation intention. It is usually set between 0.6 and 0.8, depending on the system's sensitivity to the recommendation intensity and the changing characteristics during crop growth. In this embodiment, it is set to 0.7, which can effectively distinguish the intervention needs with strong recommendation intentions and ensure that the system can respond to more urgent situations in a timely manner.

[0165] The preset height change fluctuation threshold refers to a standard used to determine whether the crop growth changes are too drastic. It depends on the growth characteristics of the crop and the impact of environmental changes. It is usually set between 0.1 and 0.3. In this embodiment, it is set to 0.2. It can effectively capture abnormal fluctuations in crop growth, thereby helping the system to promptly identify situations that require intervention, ensuring that agricultural input intervention can be carried out in a timely manner when large fluctuations in crop growth occur.

[0166] When the recommendation intent strength exceeds a preset strength threshold, the system first calculates the absolute value of the difference in crop growth height within adjacent preset time windows to obtain several height change rates. Next, it calculates the standard deviation of these height change rates to obtain a height change fluctuation value. Finally, if this fluctuation value exceeds a preset height change fluctuation threshold, the response level is determined to be the priority level.

[0167] By analyzing changes in crop height, the system determines crop growth status and prioritizes high-need agricultural interventions in situations of significant growth fluctuation. By setting preset intensity thresholds and height fluctuation thresholds, the system intelligently identifies situations requiring immediate intervention, avoiding wasted resources and improving the efficiency of agricultural intervention responses. Furthermore, the logical correlation between data—the correlation between crop growth fluctuations and intervention needs—is reflected in the analysis of height change rate and volatility, further enhancing the accuracy and responsiveness of the recommendation system.

[0168] Specifically, the adjustment module includes:

[0169] a recording unit, configured to record a current timestamp to form an initial time point when the priority level occurs more than a preset threshold number of times;

[0170] a number acquisition unit connected to the recording unit, configured to acquire the number of times within a time period starting from the initial time point to each time stamp within the preset adjustment time period, to form a plurality of times;

[0171] a frequency fluctuation calculation unit, connected to the frequency acquisition unit, for calculating the standard deviation of all the frequencies to form a frequency fluctuation value;

[0172] An adjusting unit is used to adjust the severity threshold according to the number fluctuation value to form an adjusted threshold.

[0173] The first recording unit records a current timestamp to form an initial time point when the priority level appears more than a preset number threshold. Then, the number obtaining unit calculates time intervals of timestamps within a preset adjustment duration from the initial time point to form a plurality of numbers. The number fluctuation calculating unit calculates a standard deviation of the numbers to obtain a number fluctuation value. Finally, the adjustment unit adjusts the severity threshold based on the number fluctuation value to form a new adjustment threshold for updating in a subsequent intervention determination.

[0174] The system can dynamically adjust the severity threshold based on the fluctuation of the frequency of the priority level. If the priority level appears frequently and fluctuates greatly within a period of time, it indicates that the intervention demand may intensify. The system can adjust the threshold to strengthen the response, so that the system can flexibly respond to various changing intervention demands, thereby achieving more accurate agricultural intervention, optimizing crop growth management, and effectively avoiding resource waste or over-intervention.

[0175] Specifically, the adjustment unit comprises:

[0176] The number fluctuation comparison subunit is configured to compare the number fluctuation value with a preset number fluctuation threshold to form a number fluctuation comparison result.

[0177] The adjustment subunit is connected with the number comparison subunit and configured to increase the severity threshold based on a relative deviation between the number fluctuation value and the preset number fluctuation threshold and a preset adjustment coefficient to form an adjustment threshold when the number fluctuation comparison result indicates that the number fluctuation value is greater than the preset number fluctuation threshold.

[0178] The preset adjustment coefficient is a proportional factor for adjusting the severity threshold, which is used to control the sensitivity of the adjustment amplitude and depends on the balance between response speed and stability in the actual application scenario. If the system needs to be more sensitive to fluctuation changes, a higher adjustment coefficient can be set. Conversely, to avoid frequent adjustments, a lower value can be set, usually between 0.1 and 0.5. In this embodiment, the adjustment coefficient is set to 0.3, which can ensure that the adjustment of the severity threshold is responsive and not frequently changed due to slight fluctuations, thereby improving the robustness and accuracy of the system in long-term operation.

[0179] The number fluctuation comparison subunit first compares the number fluctuation value within the preset adjustment duration with the preset number fluctuation threshold to form a number fluctuation comparison result. When the comparison result indicates that the number fluctuation value is greater than the preset threshold, the adjustment subunit adjusts the severity threshold based on the relative deviation between the two and the preset adjustment coefficient to obtain an updated adjustment threshold, thereby dynamically optimizing the event determination standard.

[0180] By comparing the frequency fluctuation value with the preset frequency fluctuation threshold, and combining the relative deviation and the preset adjustment coefficient for parameter adjustment, the severity threshold can automatically respond to changes according to the fluctuation of actual intervention requirements. This mechanism enhances the system's perception of abnormal frequency changes, realizes adaptive optimization of intervention sensitivity, ensures timely increase of the response threshold in the case of abnormal frequent fluctuations, reduces false positives and intervention redundancy, and improves the scientificity and stability of the intervention strategy.

[0181] In particular, the output module comprises:

[0182] The retrieval unit is configured to retrieve a plurality of agricultural products from the preset product library according to the re-determined recommended interaction type, to obtain a plurality of candidate products.

[0183] The usage change rate calculation unit is connected with the retrieval unit and configured to calculate the difference between the historical usage frequencies of each adjacent preset time window within a preset historical time length for each candidate product, to obtain a plurality of usage change rates.

[0184] The usage change fluctuation calculation unit is connected with the usage change rate calculation unit and configured to calculate the standard deviation of all the usage change rates, to obtain a usage change fluctuation value.

[0185] The sorting unit is connected with the usage change fluctuation calculation unit and configured to sort the usage change fluctuation values of all the candidate products from small to large, to form a product list.

[0186] The output unit is connected with the sorting unit and configured to recommend the candidate products in the top preset ranking number of the product list, to output the recommendation result.

[0187] The preset historical time length refers to the historical time range for calculating the usage change rate of the candidate product, which determines the time span within which the system analyzes the change trend of the product usage frequency. It depends on the growth cycle of crops, the typical frequency change rhythm of agricultural products, and user behavior characteristics, etc. It is usually set between 7 days and 30 days. In this embodiment, it is set to 14 days, which can balance the analysis needs of short-term fluctuations and long-term trends, effectively capture the change pattern of product usage frequency, and thus improve the accuracy of usage change fluctuation value calculation, supporting more scientific and reasonable product sorting and recommendation.

[0188] The output module filters a plurality of candidate agricultural products meeting the conditions from the preset product library through the retrieval unit based on the redetermined recommended interaction type. Then, the change rate calculation unit calculates the difference of the historical use frequency of the candidate products in the past preset historical length and the adjacent preset time window to obtain the use change rate. The use change fluctuation calculation unit calculates the standard deviation of all use change rates, i.e., the use change fluctuation value, which reflects the stability of the use frequency of each product. The sorting unit sorts the candidate products from small to large according to the fluctuation value, and preferentially recommends the products with smaller fluctuation. Finally, the output unit outputs the products ranked in the preset number as the recommended result.

[0189] By associating multiple data such as the recommended interaction type, the historical use frequency, the change rate and the fluctuation thereof, the recommended result is not only based on the demand type matching, but also considers the stability and trend of the historical use of the agricultural products, thereby improving the scientificity, adaptability and credibility of the recommended products, which helps the user to more reasonably select the agricultural products in different situations.

[0190] The above only describes the preferred embodiments of the present application and is not used to limit the present application; for those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An agricultural product recommendation system based on a data model, characterized in that: include: The acquisition module is used to obtain in real time the irrigation frequency of all crops, the growth height of all crops, the number of fields when entering description content, keyword density, input speed, and the historical usage frequency of each agricultural product in the preset product library within a preset recording period from the start of target input; a prediction module connected to the acquisition module, configured to predict a severity threshold based on a preset agricultural input model, the irrigation frequency of all the crops, and the growth height of all the crops; an event determination module, connected to the acquisition module and the prediction module respectively, for determining the occurrence of an agricultural input intervention demand event based on the number of fields, the irrigation frequency of all crops, and the severity threshold, and forming an intervention determination result; a type determination module, connected to the acquisition module and the event determination module respectively, for determining a recommended interaction type for the current target based on the intervention determination result and the input speed; a level determination module, connected to the acquisition module and the type determination module respectively, for determining the priority level of this recommendation according to the recommended interaction type, the crop growth height, the keyword density, and the number of fields; an adjustment module, connected to the level determination module and the prediction module respectively, for adjusting the severity threshold according to the number of occurrences of the priority level within a preset adjustment time period to form an adjustment threshold; an output module connected to the type determination module, configured to output a recommendation result based on the recommended interaction type re-determined by the adjustment threshold and the historical usage frequency; The event determination module includes: an average calculation unit, configured to calculate an average of the irrigation frequencies of all the crops to form an average irrigation frequency; a first normalization unit connected to the mean value calculation unit, configured to perform normalization processing on the average irrigation frequency to obtain a normalized irrigation frequency, and to perform normalization processing on the field quantity to obtain a normalized field quantity; an intervention index calculation unit connected to the first normalization unit, configured to perform a weighted sum calculation on the normalized irrigation frequency, the preset irrigation frequency weight, the normalized number of fields, and the preset field number weight to obtain an intervention index; an event determination unit connected to the intervention index calculation unit, configured to determine the occurrence of the agricultural input intervention demand event based on the intervention index and the severity threshold, and form the intervention determination result; The event determination unit includes: A first determination subunit is configured to determine that the agricultural input intervention demand event occurs when the intervention index is greater than the severity threshold, and form the intervention determination result; a frequency change calculation subunit, connected to the first determination subunit, for calculating, when the intervention index is less than or equal to the severity threshold, the absolute value of the difference between the crop irrigation frequencies in all adjacent preset time windows within the preset recording duration to form a plurality of frequency change rates; a change fluctuation calculation subunit, connected to the frequency change calculation subunit, for calculating the standard deviation of all the frequency change rates to form a change rate fluctuation value; The second determination subunit is connected to the change fluctuation calculation subunit and is used to determine that the agricultural input intervention demand event occurs when the change rate fluctuation value is greater than a preset change rate fluctuation threshold, and form the intervention determination result.

2. The agricultural product recommendation system based on data model according to claim 1 is characterized in that: The type determination module includes: a speed comparison unit, configured to compare the input speed with a preset speed threshold when forming the intervention determination result, to form a speed comparison result; A type determination unit is connected to the speed comparison unit and is configured to determine that the recommended interaction type is an emergency intervention type when the speed comparison result shows that the input speed is greater than the preset speed threshold.

3. The agricultural product recommendation system based on data model according to claim 2 is characterized in that: The level determination module includes: a strength determination unit, configured to determine the strength of the recommendation intention according to the keyword density and the number of fields when the emergency intervention type is determined; A priority level determination unit is connected to the strength determination unit and is used to determine the response level of the current recommendation as the priority level according to the strength of the recommendation intention and the growth height of the crop.

4. The agricultural product recommendation system based on data model according to claim 3 is characterized in that: The strength determination unit includes: a density fluctuation calculation subunit, configured to calculate the standard deviation of the density of all the keywords within the preset recording time to obtain a density fluctuation value; A word count fluctuation calculation subunit, configured to calculate a standard deviation of the number of all fields within the preset recording time to obtain a word count fluctuation value; a normalization subunit, connected to the density fluctuation calculation subunit and the word count fluctuation calculation subunit, respectively, for normalizing the density fluctuation value to obtain a density normalized fluctuation value, and normalizing the word count fluctuation value to obtain a word count normalized fluctuation value; a consistency calculation subunit, connected to the normalization subunit, for calculating a correlation coefficient between the density normalized fluctuation value and the word count normalized fluctuation value to obtain a fluctuation consistency; The strength determination subunit is connected to the consistency calculation subunit and is used to calculate the relative deviation between the fluctuation consistency and the preset standard consistency when the fluctuation consistency is greater than the preset standard consistency, so as to determine the strength of the recommendation intention.

5. The agricultural product recommendation system based on data model according to claim 4 is characterized in that: The priority determination unit includes: a height change calculation subunit, configured to calculate, when the strength of the recommendation intention is greater than a preset strength threshold, the absolute value of the difference between the crop growth heights in all adjacent preset time windows to obtain a plurality of height change rates; an altitude change fluctuation calculation subunit, connected to the altitude change calculation subunit, for calculating the standard deviation of all the altitude change rates to form an altitude change fluctuation value; The priority level determination subunit is connected to the altitude change fluctuation calculation subunit and is used to determine that the response level is a priority level when the altitude change fluctuation value is greater than a preset altitude change fluctuation threshold.

6. The agricultural product recommendation system based on data model according to claim 5 is characterized in that: The adjustment module includes: a recording unit, configured to record a current timestamp to form an initial time point when the priority level occurs more than a preset threshold number of times; a number acquisition unit connected to the recording unit, configured to acquire the number of times within a time period starting from the initial time point to each time stamp within the preset adjustment time period, to form a plurality of times; a frequency fluctuation calculation unit, connected to the frequency acquisition unit, for calculating the standard deviation of all the frequencies to form a frequency fluctuation value; An adjusting unit is used to adjust the severity threshold according to the number fluctuation value to form an adjusted threshold.

7. The agricultural product recommendation system based on data model according to claim 6 is characterized in that: The adjustment unit includes: A number fluctuation comparison subunit is used to compare the number fluctuation value with a preset number fluctuation threshold to form a number fluctuation comparison result; An adjustment subunit is connected to the number comparison subunit and is used to increase the severity threshold according to the relative deviation between the number fluctuation value and the preset number fluctuation threshold and the preset adjustment coefficient to form an adjustment threshold when the number fluctuation comparison result is that the number fluctuation value is greater than the preset number fluctuation threshold.

8. The agricultural product recommendation system based on data model according to claim 7 is characterized in that: The output module includes: a retrieval unit, configured to retrieve a plurality of agricultural products from the preset product library according to the re-determined recommendation interaction type, so as to obtain a plurality of candidate products; a usage change rate calculation unit connected to the retrieval unit, for calculating the difference in the historical usage frequency of each candidate product in each adjacent preset time window within a preset historical duration to obtain a plurality of usage change rates; a usage change fluctuation calculation unit connected to the usage change rate calculation unit, for calculating a standard deviation of all the usage change rates to obtain a usage change fluctuation value; a sorting unit connected to the usage change fluctuation calculation unit, for sorting the usage change fluctuation values ​​of all the candidate products from small to large to form a product list; An output unit is connected to the ranking unit and is used to recommend the candidate products with the first preset ranking number in the product list to output the recommendation result.

Images