Marketing recommendation quotation method and device based on large model, equipment and medium

Abstract

The invention relates to the field of intelligent decision making, finance and medical health, and discloses a marketing recommendation quotation method, device, equipment and medium based on a large model, and the method comprises the steps: obtaining first feature data from multiple channels based on a customer name, and inputting the first feature data and customer personalized demands into a plurality of preset large models; the output quotation scheme is an accurate quotation scheme generated by combining the industry trend, the market dynamic state and the customer demand, then the preset rule engine and the manual terminal are used for auditing, and finally the target quotation scheme with the highest return rate is determined, so that the target quotation scheme also meets the company management and control requirements. The method comprises the steps of obtaining first feature data based on a customer name; inputting the quotation scheme and the personalized requirements of the customer into a plurality of preset large models, and outputting the quotation scheme; auditing the quotation scheme by using a preset rule engine; if the management and control requirements are met, the quotation scheme is sent to a manual terminal to cooperatively confirm whether the quotation scheme is feasible; and finally, screening out a target quotation scheme.

Description

Technical Field

[0001] The present invention relates to the fields of intelligent decision-making, finance and medical health, and in particular to a marketing recommendation quotation method, device, equipment and medium based on a large model. Background Art

[0002] The current marketing market is highly competitive, and the quotation proposal to the customer often directly determines whether the customer can be successfully acquired.

[0003] In related technologies, quotation schemes usually rely on a single data source, such as historical transaction data or public market data, and the generated quotation schemes are often not recognized by customers.

[0004] Therefore, how to generate more accurate quotation plans for customers and improve customer recognition has become a technical problem that needs to be solved urgently by technical personnel in this field. Summary of the Invention

[0005] The present invention provides a marketing recommendation quotation method, device, equipment and medium based on a large model to solve the technical problem that the generated quotation scheme is inaccurate and cannot be recognized by customers.

[0006] First, a marketing recommendation quotation method based on a large model is provided, comprising:

[0007] Receive the customer name entered by the account manager;

[0008] Based on the customer name, first feature data is obtained from multiple channels, where the first feature data includes a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price;

[0009] Inputting the first feature data and the customer's personalized needs into a plurality of preset large models to output quotation solutions corresponding to the preset large models;

[0010] Using a preset rule engine, review whether the quotation proposal meets the control requirements to obtain a first review result;

[0011] If the first review result is that it meets the control requirements, it is sent to the manual end, so that the manual reviewer can collaboratively confirm whether the quotation scheme that meets the control requirements is feasible through the manual end to obtain the second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected;

[0012] Filter out the bidding proposal with the highest rate of return as the target bidding proposal.

[0013] In a second aspect, a marketing recommendation and quotation device based on a large model is provided, comprising:

[0014] A receiving unit, used to receive the customer name input by the account manager;

[0015] A first acquisition unit is configured to acquire first characteristic data from multiple channels based on the customer name, the first characteristic data including a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price;

[0016] An input unit, configured to input the first feature data and the customer's personalized needs into a plurality of preset large models, so as to output quotation solutions corresponding to the preset large models;

[0017] A judgment unit, configured to use a preset rule engine to review whether the quotation proposal meets the management and control requirements to obtain a first review result;

[0018] a confirmation unit configured to send the first review result to a manual end if it is in compliance with the control requirements, so that the manual reviewer can collaboratively confirm through the manual end whether the quotation scheme that meets the control requirements is feasible to obtain a second review result; and determine the quotation scheme that is feasible according to the second review result as the quotation scheme to be selected;

[0019] The screening unit is used to screen out the quotation scheme with the highest rate of return as the target quotation scheme.

[0020] In a third aspect, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned marketing recommendation quotation method based on a large model are implemented.

[0021] In a fourth aspect, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the steps of the above-mentioned marketing recommendation quotation method based on a large model are implemented.

[0022] In the solution implemented by the above-mentioned marketing recommendation quotation method, device, equipment and medium based on the big model, the customer name input by the customer manager can be received; based on the customer name, the first characteristic data is obtained from multiple channels, and the first characteristic data includes the first industry growth rate, the first GDP, the first supply chain procurement amount and the first competitor price; the first characteristic data and the customer's personalized needs are input into multiple preset big models to output a quotation scheme corresponding to the preset big model; using the preset rule engine, the quotation scheme is reviewed to see whether it meets the management and control requirements to obtain a first review result; if the first review result is that it meets the management and control requirements, it is sent to the manual end so that the manual reviewer can collaboratively confirm through the manual end whether the quotation scheme that meets the management and control requirements is feasible to obtain a second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected; and the quotation scheme to be selected with the highest return rate is screened out as the target quotation scheme. In the present invention, first feature data including multiple data is obtained from multiple channels based on the customer name, and at the same time, it is input into multiple preset large models together with the customer's personalized needs to output multiple quotation schemes. The quotation scheme output in this way is an accurate quotation scheme generated by combining industry trends, market dynamics and customer needs. Then, a preset rule engine and manual review are used to determine the quotation scheme to be selected that meets the management and control requirements, and finally the quotation scheme to be selected with the highest return rate is screened out as the target quotation scheme. In this way, the target quotation scheme finally determined not only meets the customer's personalized needs, but also meets the company's management and control requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0024] Figure 1 This is a schematic diagram of an application environment of a marketing recommendation quotation method based on a large model in one embodiment of the present invention;

[0025] Figure 2 A schematic diagram of a flow chart of a marketing recommendation quotation method based on a large model provided by an embodiment of the present invention;

[0026] Figure 3 The following is a schematic diagram illustrating the structure of a marketing recommendation and quotation device based on a large model according to some embodiments;

[0027] Figure 4 is a structural diagram of a computer device in one embodiment of the present invention;

[0028] Figure 5FIG. 2 is another structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0030] Competition in the current marketing market is fierce, and the price quote provided to customers often directly determines successful customer acquisition. Related technologies often rely on a single data source, such as historical transaction data or publicly available market data, to generate quotes that often fail to meet customer satisfaction. Therefore, how to generate more accurate quotes for customers and improve their acceptance has become a pressing technical challenge for those skilled in the art.

[0031] The marketing recommendation quotation method based on the large model provided by the embodiment of the present invention can be applied in the following aspects: Figure 1 In an application environment, a customer manager inputs a customer name through a customer manager terminal. The service terminal receives the customer name input by the customer manager; based on the customer name, first feature data is obtained from multiple channels, the first feature data including a first industry growth rate, a first GDP, a first supply chain purchase amount, and a first competitor price; the first feature data and the customer's personalized needs are input into multiple preset large models to output a quotation scheme corresponding to the preset large model; a preset rule engine is used to review whether the quotation scheme meets the control requirements to obtain a first review result; if the first review result is that it meets the control requirements, it is sent to the manual terminal so that the manual reviewer can collaboratively confirm whether the quotation scheme that meets the control requirements is feasible through the manual terminal to obtain a second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected; and the quotation scheme to be selected with the highest return rate is screened out as the target quotation scheme. In the present invention, first feature data including multiple data is obtained from multiple channels based on the customer name, and at the same time, it is input into multiple preset large models together with the customer's personalized needs to output multiple quotation schemes. The quotation scheme output in this way is an accurate quotation scheme generated by combining industry trends, market dynamics and customer needs. Then, a preset rule engine and manual review are used to determine the quotation scheme to be selected that meets the management and control requirements, and finally the quotation scheme to be selected with the highest return rate is screened out as the target quotation scheme. In this way, the target quotation scheme finally determined not only meets the customer's personalized needs, but also meets the company's management and control requirements.

[0032] The account manager terminal can be, but is not limited to, various personal computers, laptops, smartphones, tablet computers, and portable wearable devices. The server terminal can be implemented as a standalone server or a server cluster consisting of multiple servers. The present invention is described in detail below using specific embodiments.

[0033] See also Figure 2 As shown, Figure 2 A flowchart of a marketing recommendation quotation method based on a large model provided in an embodiment of the present invention includes the following steps S100-S600.

[0034] S100: Receive the customer name input by the account manager.

[0035] In the embodiment of the present application, the account manager can enter the customer name in the quotation assistant. Specifically, the quotation assistant is an application program, and the customer name can be entered on the interface of the quotation assistant.

[0036] The large-scale model-based marketing recommendation and quotation method in the embodiments of this application can be applied in multiple fields, including finance and healthcare. For example, in the financial field, insurance quotes can be provided. In the healthcare field, full-service physical examination quotes can be provided.

[0037] S200. Based on the customer name, obtain first characteristic data from multiple channels, where the first characteristic data includes a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price.

[0038] In an embodiment of the present application, a quotation assistant can be used to obtain first feature data from multiple channels based on the customer name.

[0039] In some embodiments, the step of obtaining the first feature data from multiple channels is the same as the step of obtaining the second feature data from multiple channels below, and will not be repeated here. The difference is that the first feature data is only data related to the customer's field corresponding to the customer name, and the second feature data is used to train the large model, so different second feature data can be data related to all fields, so that the large model can be used in different fields.

[0040] S300: Input the first feature data and the customer's personalized needs into a plurality of preset large models to output quotation plans corresponding to the preset large models.

[0041] In this embodiment of the present application, the customer's personalized needs are the various personalized needs of the customer reported by the account manager during follow-up. These needs are input into the macro model along with the first feature data to obtain a more accurate quotation solution. In this embodiment of the present application, the first feature data and the customer's personalized needs are input into multiple preset macro models, so that each preset macro model can output a quotation solution. This allows the optimal quotation solution to be subsequently determined from the multiple quotation solutions, thereby improving the accuracy of the quotation solution.

[0042] In the embodiment of the present application, the customer's personalized needs can be input on the quotation assistant interface.

[0043] In some embodiments, the method further includes determining a preset macro model, specifically including:

[0044] Second characteristic data is obtained from multiple channels, where the second characteristic data includes a second industry growth rate, a second GDP, a second supply chain procurement amount, and a second competitor product price.

[0045] In some embodiments, the second feature data may also include: customer-related parameters and customer feedback and optimization parameters. Customer-related parameters include: 1. Customer credit rating, which is used to assess the customer's payment ability and risk, and influences whether to grant preferential payment terms or discounts. Data sources: customer historical transaction records, credit rating agencies, bank credit information, etc. 2. Customer historical purchasing behavior, including purchase frequency, purchase volume, and purchasing preferences, which helps determine customer value and formulate long-term cooperation strategies. For example, high-frequency, large-volume customers may enjoy tiered pricing discounts. 3. Customer scale: Company size (such as number of employees, annual revenue), etc., which influences price sensitivity and negotiation room. Customer feedback and optimization parameters include: 1. Customer acceptance of the quotation: Assessing the market acceptance of the quotation through historical transaction data and customer feedback. 2. Customer negotiation records and preferences: Recording customer feedback during negotiations, such as preferences for price, payment method, and service content, for subsequent optimization. 3. Quote success rate and conversion rate: Optimizing the model output strategy by analyzing the customer's transaction status after the quotation is generated.

[0046] Correspondingly, the first characteristic data may also include the above-mentioned customer-related parameters and customer feedback and optimization parameters.

[0047] In some embodiments, the step of obtaining the second feature data from multiple channels includes:

[0048] Obtain multi-dimensional data from various channels, including industry data, regional data, supply chain data, competitive product data, and the company's TCR indicator (Total Cash Return Rate, calculated as: Total Revenue / Total Capital Occupied)

[0049] In the embodiments of this application, the multiple channels include industry-level data, government public platforms, and local data released by the Economic Statistics Bureau. In addition, it also includes obtaining data and information related to the company's cooperative supply chain and other competitive product-related information through other means such as technology or public purchase.

[0050] In the examples of this application, an efficient and accurate intelligent quotation generation system is implemented by integrating multi-dimensional data, introducing large models for in-depth analysis, and combining automated and intelligent processing processes. From data collection, processing, analysis, to quotation generation, every step of the system has been carefully optimized and designed to ensure the system's efficiency, scalability, and security. This solution enables enterprises to quickly generate targeted quotation solutions, improve market competitiveness and customer satisfaction, while optimizing resource allocation and reducing operating costs.

[0051] All the multi-dimensional data are deduplicated, abnormal data are removed, and missing data are supplemented to obtain processed multi-dimensional data.

[0052] In some embodiments, the step of supplementing the missing data includes:

[0053] Missing data and the type of missing data are determined.

[0054] In this embodiment, missing data in multi-dimensional data can be determined by using data cleaning tools (such as Pandas).

[0055] In some embodiments, missing data may be industry data, such as missing GDP growth rates or industry market size data for certain regions. Missing data may be supply chain data, such as missing procurement costs or lead times for a supplier. Missing data may also be competitor data, such as missing market share or product pricing data for certain competitors. Missing data may also be company TCR metrics: for example, missing return on equity (ROE) or transaction cost (TC) data for certain quarters. Missing data may also be product information: for example, missing cost or profit margin data for certain products.

[0056] In this embodiment, the types of missing data include: missing at random (MAR), missing not at random (MNAR) and missing completely at random (MCAR).

[0057] In some embodiments, methods for determining the type of missing data include observing missing data distribution, statistical testing, business logic analysis, and expert interviews or customer manager feedback.

[0058] Specifically, initially observe the missingness distribution: Use visualization tools (such as heat maps and missingness matrices) to determine which variables are missing and whether the missingness is concentrated in certain customers or regions. If the missingness is evenly distributed, it may be MCAR; if it is concentrated in certain variables or customer groups, it may be MAR or MNAR.

[0059] Statistical Tests: For categorical variables, use the chi-square test to compare the distribution of missing values ​​and non-missing values ​​on other variables. For continuous variables, use the t-test or ANOVA to compare the mean difference between missing and non-missing values. If the difference is significant, it indicates that the missing values ​​may be related to certain variables and is determined to be MAR.

[0060] Business logic analysis: Based on actual factoring scenarios, analyze whether missing data is related to customer behavior, data entry habits, system issues, etc. For example, if some customers are unwilling to disclose sensitive information and their data is missing, this is determined to be MNAR.

[0061] Expert interviews or account manager feedback: Communicate with account managers to understand the reasons for missing data. If it is found that some clients did not provide data due to high risk, it is determined to be MNAR.

[0062] Based on the type of missing data, determine the supplementation method.

[0063] In this embodiment, if the missing data is missing non-randomly, the supplementation method is determined to be a machine learning model and expert evaluation. If the missing data is missing completely at random, the supplementation method is determined to be interpolation. If the missing data is missing at random, the supplementation method is determined to be an extrapolation method using industry benchmark data and historical data, such as indicators such as industry growth rate, regional GDP, and supply chain procurement.

[0064] Based on the supplementation method, the missing data is supplemented.

[0065] The following describes how to supplement missing data using different supplementation methods.

[0066] (1) Interpolation method

[0067] Interpolation is a common method for supplementing missing data and is suitable for time series data or data with continuous characteristics.

[0068] Linear interpolation: Applicable to time series data. For example, if the GDP growth rate of a certain region is missing, linear interpolation can be performed using the GDP growth rates of two previous and next time points.

[0069] For example: The GDP growth rate of a region is 5% in 2021 and 6% in 2023. The data for 2022 is missing. The GDP growth rate in 2022 can be inferred to be 5.5% through linear interpolation.

[0070] Mean / Median Interpolation: Applicable to non-time series data. For example, if the procurement cost of a supplier is missing, it can be supplemented by the mean or median procurement cost of other suppliers in the same industry.

[0071] For example: If the purchase cost of a supplier is missing, it can be supplemented by the average purchase cost of other suppliers in the same industry (such as 1 million yuan).

[0072] (2) Machine Learning Model Prediction

[0073] For complex missing data, machine learning models can be used for prediction.

[0074] Regression model: For example, use linear regression or random forest regression models to predict missing supply chain procurement costs based on other relevant features (such as industry growth rate, regional economic indicators).

[0075] For example, if the procurement cost of a certain supplier is missing, a regression model can be trained to make predictions based on the procurement costs of other suppliers, industry growth rates, regional economic indicators and other characteristics.

[0076] Cluster analysis: For example, use the K-Means clustering algorithm to classify the categories where the missing data are located, and then use the mean or median within the category to fill in the missing values.

[0077] For example: if the profit margin of a product is missing, it can be classified into similar products through cluster analysis and supplemented with the average profit margin of similar products.

[0078] (3) Industry benchmark data

[0079] For some missing data, you can refer to the industry's publicly available benchmark data for supplementation.

[0080] Industry report: For example, if the GDP growth rate of a certain region is missing, it can be supplemented by the average GDP growth rate of the region in the industry report.

[0081] Market research data: For example, if a competitor’s market share is missing, it can be supplemented by the average market share of the industry in the market research report.

[0082] (4) Historical data extrapolation

[0083] For data with time series characteristics, historical data can be used for extrapolation.

[0084] Trend extrapolation: For example, if a company’s return on equity (ROE) is missing, it can be extrapolated based on the trend of historical ROE data.

[0085] For example: A company's ROE in 2021 is 10%, its ROE in 2022 is 12%, and the data for 2023 is missing. It can be inferred through linear trend that the ROE in 2023 is 14%.

[0086] Seasonal adjustment: For example, if the sales data of a product has seasonal characteristics, it can be supplemented by historical seasonal data.

[0087] For example, if a product has higher sales in summer and lower sales in winter, the missing winter sales can be inferred using historical seasonal data.

[0088] (5) Expert evaluation

[0089] For some missing data that is difficult to supplement through models or historical data, an assessment can be conducted based on the experience of business experts.

[0090] Expert scoring method: For example, if the profit margin of a product is missing, it can be evaluated through the experience of business experts, combined with market research data and industry trends, to give a reasonable profit margin value.

[0091] Delphi method: Through multiple rounds of expert opinions, a reasonable supplementary value is obtained.

[0092] The above is the specific content of using different methods to supplement missing data.

[0093] In some embodiments, the method further comprises verifying the supplementary data. Specifically, the rationality of the supplementary data is verified by statistical analysis (such as mean, variance) or visualization tools (such as box plot).

[0094] The processed multi-dimensional data is extracted to obtain second feature data.

[0095] In this embodiment, the second characteristic data is mainly data related to the quotation.

[0096] Using the second feature data, multiple large models are trained to obtain corresponding multiple preset large models, wherein the control requirements are embedded as constraints into the training objective function of the large model.

[0097] In this embodiment of the present application, the multiple large models may include a DeepSeek model, a Qwen model, and an LLM model. Control requirements are embedded as constraints in the training objective function of each large model. The large model performs in-depth analysis of the second feature data and is trained and optimized to enable the large model to understand the relevance of the second feature data and utilize the constraints to generate a quotation solution that meets the various control requirements.

[0098] S400: Using a preset rule engine, review whether the quotation proposal meets the management and control requirements to obtain a first review result.

[0099] In the embodiment of the present application, after generating a quotation plan using the preset large model, the preset rule engine is used to continue to review whether the quotation plan meets the management and control requirements, so as to ensure that the final target quotation plan meets the management and control requirements.

[0100] In the embodiment of the present application, the enterprise can dynamically modify the control requirements according to market changes, policy adjustments, internal strategy updates, etc. For example, the control requirement can be that customers with a credit rating of C are not allowed to have an account period exceeding 30 days.

[0101] In some embodiments, before executing step S400, the method further includes building a preset rule engine; specifically, the method further includes:

[0102] A rule base is set up, wherein the rule base includes executable business rules converted from the control requirements, and the executable business rules are set with corresponding priorities; based on the rule base, a preset rule engine is constructed.

[0103] The control requirements in the embodiments of this application include cost-profit management and risk management. Cost-profit management includes the cost coverage principle and TCR control. The cost coverage principle: The quotation must cover the direct and indirect costs of the product or service. TCR control: Ensure that the quotation achieves the cash return target while meeting transaction costs. Risk management includes customer credit and legal compliance control.

[0104] In the embodiments of the present application, executable business rules are executable code data. For example, a control requirement may be that customers with a credit rating of C are not allowed to have an account period exceeding 30 days. This control requirement can be converted into an executable business rule, and after assigning a corresponding priority to the executable business rule, it can be stored in a rule library. This rule library is used to build a preset rule engine.

[0105] In an embodiment of the present application, setting corresponding priorities for executable business rules can ensure that when multiple management and control requirements apply at the same time, the quotation assistant can make reasonable judgments on whether the quotation plan meets the management and control requirements in order of priority.

[0106] In some embodiments, it also includes dynamically modifying the control requirements according to market changes, policy adjustments, and internal strategy updates, and then modifying the rule base based on the dynamically modified control requirements.

[0107] In this embodiment, when customer needs or company strategy adjustments are updated in real time, the rule base and the management and control requirements of the training objective function in the preset large model used as constraints embedded in the large model are also updated in real time, thereby dynamically generating personalized quotation plans.

[0108] In some embodiments, the step of using a preset rule engine to review whether the quotation proposal complies with the management and control requirements to obtain a first review result includes:

[0109] The preset rule engine is used to sort the executable business rules from high to low according to the priority in the rule library, and determine whether the quotation solution complies with the first executable business rule.

[0110] If the executable business rules are not met, the first audit result is determined to be non-compliant with the control requirements;

[0111] If it complies with the executable business rules, continue to determine whether the quotation plan complies with the executable business rules corresponding to the next priority level until it complies with all executable business rules, and determine that the first review result complies with the management and control requirements.

[0112] Exemplarily, the priority of executable business rule A is higher than the priority of executable business rule B, which is higher than the priority of executable business rule C. At this time, it is first determined whether the quotation plan complies with executable business rule A. If it does not comply with executable business rule A, the first audit result is determined to be non-compliant with management and control requirements. If it complies with executable business rule A, it is further determined whether the quotation plan complies with executable business rule B. If it does not comply with executable business rule B, the first audit result is determined to be non-compliant with management and control requirements. If it complies with executable business rule B, it is further determined whether the quotation plan complies with executable business rule C. If it does not comply with executable business rule C, the first audit result is determined to be non-compliant with management and control requirements. If it complies with executable business rule C, the first audit result is determined to be compliant with management and control requirements.

[0113] S500. If the first audit result is in compliance with the control requirements, it is sent to the manual end so that the manual reviewer can collaboratively confirm through the manual end whether the quotation scheme that meets the control requirements is feasible to obtain the second audit result; the quotation scheme that is feasible according to the second audit result is determined as the quotation scheme to be selected.

[0114] In some embodiments, there may be multiple manual terminals, and the manual terminals may be devices such as mobile phones and computers. Exemplarily, the multiple manual terminals may include a sales terminal, a financial terminal, and a legal terminal. Different manual reviewers collaboratively confirm the quotation scheme through the corresponding manual terminals. The step in which the manual reviewers collaboratively confirm whether the quotation scheme that meets the management and control requirements is feasible through the manual terminals to obtain the second audit result includes: judging whether the quotation schemes that meet the management and control requirements determined by all manual terminals are feasible. If all are feasible, the second audit result is determined to be feasible. If one of the manual terminals determines that the quotation scheme that meets the management and control requirements is not feasible, the second audit result is determined to be infeasible.

[0115] In the embodiment of the present application, there may be multiple quotation schemes that are feasible according to the second review result, and each quotation scheme that is feasible according to the second review result is determined as a quotation scheme to be selected. Subsequently, an optimal quotation scheme is screened out from the multiple quotation schemes to be selected.

[0116] In some embodiments, before sending the quotation to the manual end, the method further includes: determining whether the quotation amount in the quotation proposal exceeds a preset amount. The preset amount can be set according to actual needs.

[0117] If the amount exceeds the preset amount, it is sent to the manual end; if it does not exceed the preset amount, it automatically conducts compliance, financial and legal audits to obtain a third audit result. If the third audit result is that the automatic audit is passed, step S600 is executed.

[0118] In this embodiment, the quotation proposals with larger quotation amounts (i.e., exceeding the preset amount) are further reviewed manually, while the quotation proposals with smaller quotation amounts are automatically reviewed without manual review.

[0119] S600: Filter out the bidding proposal with the highest rate of return as the target bidding proposal.

[0120] The target quotation scheme in the embodiment of the present application not only meets the customer's personalization, but also meets the company's profit requirements.

[0121] In some embodiments, the target quotation proposal is further transmitted to the user terminal. Prior to transmitting the target quotation proposal to the user terminal, the process further includes determining whether the target quotation proposal exceeds a monetary threshold. Exemplarily, the monetary threshold is 1 million. If the target quotation proposal exceeds the monetary threshold, the target quotation proposal is transmitted to the sales manager terminal and the financial director terminal, respectively, for approval by the sales manager via the sales manager terminal and the financial director via the financial director terminal. If both parties approve the target quotation proposal, the target quotation proposal is transmitted to the user terminal. The financial director terminal and the sales manager terminal can be devices such as mobile phones and computers.

[0122] In some embodiments, the method further includes: if the first audit result is not in compliance with the control requirements, feeding back the quotation proposal that does not meet the control requirements and the first audit result to the preset large model to optimize the preset large model.

[0123] In this embodiment, the preset large model also adopts a reinforcement learning mechanism. Through the reinforcement learning method, the model obtains positive and negative feedback based on whether it meets the management and control requirements, thereby optimizing the generation strategy.

[0124] In some embodiments, the method further includes: if the second audit result is infeasible, feeding back the infeasible quotation plan and the second audit result to the preset large model to optimize the preset large model.

[0125] In some embodiments, the method further includes: if the third audit result is that the automatic audit fails, feeding back the quotation plan that fails the automatic audit and the third audit result to the preset large model to optimize the preset large model.

[0126] In some embodiments, the method further includes: upon receiving a usage result of the target quotation solution, feeding the usage result and the target quotation solution back to the preset macro model to optimize the preset macro model. The usage result includes whether the customer accepts and whether the customer modifies the solution.

[0127] It can be seen that in the above scheme, the first feature data including multiple data is obtained from multiple channels based on the customer name, and at the same time, it is input into multiple preset large models together with the customer's personalized needs to output multiple quotation schemes. The quotation scheme output in this way is an accurate quotation scheme generated by combining industry trends, market dynamics and customer needs. Then, the preset rule engine and manual review are used to determine the quotation scheme to be selected that meets the management and control requirements, and finally the quotation scheme to be selected with the highest return rate is selected as the target quotation scheme. In this way, the target quotation scheme finally determined not only meets the customer's personalized needs, but also meets the company's management and control requirements.

[0128] It should be understood that the order of execution of the steps in the above embodiments does not necessarily mean the order of execution. The order of execution of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0129] In one embodiment, a marketing recommendation quotation device based on a large model is provided, and the marketing recommendation quotation device based on a large model corresponds one-to-one to the marketing recommendation quotation method based on a large model in the above embodiment. Figure 3As shown, the marketing recommendation quotation device based on the large model includes: a receiving unit 301, a first obtaining unit 302, an input unit 303, a judgment unit 304, a confirmation unit 305 and a screening unit 306. The functional modules are described in detail as follows:

[0130] A receiving unit, used to receive the customer name input by the account manager;

[0131] A first acquisition unit is configured to acquire first characteristic data from multiple channels based on the customer name, the first characteristic data including a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price;

[0132] An input unit, configured to input the first feature data and the customer's personalized needs into a plurality of preset large models, so as to output quotation solutions corresponding to the preset large models;

[0133] A judgment unit, configured to use a preset rule engine to review whether the quotation proposal meets the management and control requirements to obtain a first review result;

[0134] a confirmation unit configured to send the first review result to a manual end if it is in compliance with the control requirements, so that the manual reviewer can collaboratively confirm through the manual end whether the quotation scheme that meets the control requirements is feasible to obtain a second review result; and determine the quotation scheme that is feasible according to the second review result as the quotation scheme to be selected;

[0135] The screening unit is used to screen out the quotation scheme with the highest rate of return as the target quotation scheme.

[0136] In some embodiments, the apparatus further comprises:

[0137] A setting unit, configured to set a rule base, wherein the rule base includes executable business rules converted from the control requirements, and the executable business rules are set with corresponding priorities;

[0138] A construction unit is used to construct a preset rule engine based on the rule base.

[0139] In some embodiments, the judgment unit specifically includes:

[0140] The review unit is configured to review whether the quotation proposal complies with the management and control requirements by using a preset rule engine to obtain a first review result, including the following steps:

[0141] The sorting unit is used to use the preset rule engine to sort the executable business rules from high to low according to the priority in the rule library, and determine whether the quotation plan meets the first executable business rule.

[0142] If the executable business rules are not met, the first audit result is determined to be non-compliant with the control requirements;

[0143] If it complies with the executable business rules, continue to determine whether the quotation plan complies with the executable business rules corresponding to the next priority level until it complies with all executable business rules, and determine that the first review result complies with the management and control requirements.

[0144] In some embodiments, the apparatus further comprises:

[0145] A second acquisition unit is configured to acquire second characteristic data from multiple channels, wherein the second characteristic data includes a second industry growth rate, a second GDP, a second supply chain procurement amount, and a second competitor product price;

[0146] A training unit is used to use the second feature data to train multiple large models to obtain corresponding multiple preset large models, wherein the control requirements are embedded as constraints into the training objective function of the large model.

[0147] In some embodiments, the second acquiring unit specifically includes:

[0148] A third acquisition unit is configured to acquire multi-dimensional data from multiple channels, wherein the multi-dimensional data includes industry data, regional data, supply chain data, competitor data, and the company's TCR indicator;

[0149] a processing unit, configured to remove duplicate data, remove abnormal data, and supplement missing data from all the multi-dimensional data to obtain processed multi-dimensional data;

[0150] The extraction unit is used to extract the processed multi-dimensional data to obtain second feature data.

[0151] In some embodiments, the processing unit is configured to:

[0152] a first determining unit, configured to determine the type of the missing data;

[0153] a second determining unit, configured to determine a supplementing method based on the type of the missing data;

[0154] A supplementing unit is used to supplement the missing data based on the supplementing method.

[0155] In some embodiments, the apparatus further comprises:

[0156] An optimization unit is configured to feed back the quotation proposal that does not meet the management and control requirements and the first audit result to the preset large model to optimize the preset large model if the first audit result is that the quotation proposal does not meet the management and control requirements.

[0157] The present invention provides a marketing recommendation quotation device based on a large model, which obtains first feature data including multiple data from multiple channels based on the customer name, and inputs the first feature data and the customer's personalized needs into multiple preset large models to output multiple quotation schemes. The quotation scheme output in this way is an accurate quotation scheme generated by combining industry trends, market dynamics and customer needs, and then uses a preset rule engine and manual review to determine the quotation scheme to be selected that meets the management and control requirements, and finally screens out the quotation scheme to be selected with the highest return rate as the target quotation scheme. In this way, the target quotation scheme finally determined not only meets the customer's personalized needs, but also meets the company's management and control requirements.

[0158] The specific limitations of the large-scale model-based marketing recommendation and quotation device can be found in the limitations of the large-scale model-based marketing recommendation and quotation method described above and will not be repeated here. Each module in the large-scale model-based marketing recommendation and quotation device can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so that the processor can call and execute the operations corresponding to each of the modules.

[0159] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external client via a network connection. When the computer program is executed by the processor, it realizes the functions or steps on the service side of a marketing recommendation quotation method based on a large model.

[0160] In one embodiment, a computer device is provided. The computer device may be a user terminal, and its internal structure diagram may be as follows: Figure 5As shown. The computer device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external server via a network connection. When executed by the processor, the computer program implements the functions or steps on the user side of a marketing recommendation quotation method based on a large model.

[0161] In one embodiment, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed:

[0162] Receive the customer name entered by the account manager;

[0163] Based on the customer name, first feature data is obtained from multiple channels, where the first feature data includes a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price;

[0164] Inputting the first feature data and the customer's personalized needs into a plurality of preset large models to output quotation solutions corresponding to the preset large models;

[0165] Using a preset rule engine, review whether the quotation proposal meets the control requirements to obtain a first review result;

[0166] If the first review result is that it meets the control requirements, it is sent to the manual end, so that the manual reviewer can collaboratively confirm whether the quotation scheme that meets the control requirements is feasible through the manual end to obtain the second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected;

[0167] Filter out the bidding proposal with the highest rate of return as the target bidding proposal.

[0168] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented:

[0169] Receive the customer name entered by the account manager;

[0170] Based on the customer name, first feature data is obtained from multiple channels, where the first feature data includes a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price;

[0171] Inputting the first feature data and the customer's personalized needs into a plurality of preset large models to output quotation solutions corresponding to the preset large models;

[0172] Using a preset rule engine, review whether the quotation proposal meets the control requirements to obtain a first review result;

[0173] If the first review result is that it meets the control requirements, it is sent to the manual end, so that the manual reviewer can collaboratively confirm whether the quotation scheme that meets the control requirements is feasible through the manual end to obtain the second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected;

[0174] Filter out the bidding proposal with the highest rate of return as the target bidding proposal.

[0175] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or computer device can be found in the relevant descriptions of the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0176] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0177] Those skilled in the art will clearly understand that for the sake of convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0178] The non-Company software tools or components appearing in the embodiments of this application are merely examples and do not represent actual use.

[0179] The embodiments described above are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention, and should all be included in the scope of protection of the present invention.

Claims

1. A marketing recommendation quotation method based on a large model, characterized in that: include: Receive the customer name entered by the account manager; Based on the customer name, first feature data is obtained from multiple channels, where the first feature data includes a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price; Inputting the first feature data and the customer's personalized needs into a plurality of preset large models to output quotation solutions corresponding to the preset large models; Using a preset rule engine, review whether the quotation proposal meets the control requirements to obtain a first review result; If the first review result is that it meets the control requirements, it is sent to the manual end, so that the manual reviewer can collaboratively confirm whether the quotation scheme that meets the control requirements is feasible through the manual end to obtain the second review result; the quotation scheme that is feasible according to the second review result is determined as the quotation scheme to be selected; Filter out the bidding proposal with the highest rate of return as the target bidding proposal.

2. The method according to claim 1, characterized in that Also includes: Setting a rule base, wherein the rule base includes executable business rules converted from the control requirements, and the executable business rules are set with corresponding priorities; Based on the rule base, a preset rule engine is constructed.

3. The method according to claim 2, characterized in that The step of using a preset rule engine to review whether the quotation proposal meets the management and control requirements to obtain a first review result includes: The preset rule engine is used to sort the executable business rules from high to low according to the priority in the rule library, and determine whether the quotation solution complies with the first executable business rule. If the executable business rules are not met, the first audit result is determined to be non-compliant with the control requirements; If it complies with the executable business rules, continue to determine whether the quotation plan complies with the executable business rules corresponding to the next priority level until it complies with all executable business rules, and determine that the first review result complies with the management and control requirements.

4. The method according to claim 1, wherein Also includes: obtaining second characteristic data from multiple channels, the second characteristic data including a second industry growth rate, a second GDP, a second supply chain procurement amount, and a second competitor product price; Using the second feature data, multiple large models are trained to obtain corresponding multiple preset large models, wherein the control requirements are embedded as constraints into the training objective function of the large model.

5. The method according to claim 1, wherein The step of obtaining the second characteristic data from multiple channels includes: Obtain multi-dimensional data from various channels, including industry data, regional data, supply chain data, competitor data, and company TCR indicators; Removing duplicate data, removing abnormal data, and supplementing missing data from all the multi-dimensional data to obtain processed multi-dimensional data; The processed multi-dimensional data is extracted to obtain second feature data.

6. The method according to claim 5, characterized in that The step of supplementing the missing data comprises: determining the type of missing data; Determine the supplementation method based on the type of missing data; Based on the supplementation method, the missing data is supplemented.

7. The method according to claim 1, characterized in that Also includes: If the first audit result is that the management and control requirements are not met, the quotation proposal that does not meet the management and control requirements and the first audit result are fed back to the preset large model to optimize the preset large model.

8. A marketing recommendation quotation device based on a large model, characterized in that: include: A receiving unit, used to receive the customer name input by the account manager; A first acquisition unit is configured to acquire first characteristic data from multiple channels based on the customer name, the first characteristic data including a first industry growth rate, a first GDP, a first supply chain procurement amount, and a first competitor product price; An input unit, configured to input the first feature data and the customer's personalized needs into a plurality of preset large models, so as to output quotation solutions corresponding to the preset large models; A judgment unit, configured to use a preset rule engine to review whether the quotation proposal meets the management and control requirements to obtain a first review result; a confirmation unit, configured to send the first review result to a manual end if it is in compliance with the control requirements, so that the manual reviewer can collaboratively confirm through the manual end whether the quotation scheme that meets the control requirements is feasible to obtain a second review result; Determine the quotation scheme that is feasible according to the second review result as the quotation scheme to be selected; The screening unit is used to screen out the quotation scheme with the highest rate of return as the target quotation scheme.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the steps of the marketing recommendation quotation method based on a large model as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the marketing recommendation quotation method based on a large model as claimed in any one of claims 1 to 7 are implemented.

Images