Production and marketing product data processing method and device, terminal and medium

Abstract

The invention discloses a production and marketing product data processing method and device, a terminal and a medium. According to the technical scheme provided by the invention, structured data and unstructured data of a target production and marketing product are collected firstly, then a vector autoregression model and a large language model are combined, the structured data are processed by using the vector autoregression model, and the structured data are processed by using the large language model; and capturing the dynamic relationship of the multivariable time sequence data, and predicting the future production-marketing-product development quantity. A large-scale language processing model is used for processing unstructured data and extracting deep-level features for prediction, so that external factors, such as customer preferences and market trends, influencing production, marketing and product development are quantized, finally, prediction results of the two models are integrated, fusion of structured data and unstructured data is realized, and then the weight is dynamically adjusted to realize the prediction of production, marketing and product development. And more accurate production-marketing-product development quantity prediction is realized.

Description

Technical Field

[0001] The present application relates to the technical field of big data analysis and processing, and in particular to a method, device, terminal and medium for processing production and sales product data. Background Art

[0002] Customer Relationship Management (CRM) is the core information system for enterprise management and customer relations. It integrates multifunctional modules such as sales, marketing and customer service, aiming to improve the operational efficiency and customer satisfaction of enterprises. With the rapid development of information technology, enterprises are able to obtain and store more and more customer data. The effective analysis of this data relies on big data technology, which enables enterprises to extract valuable information from massive data for decision support.

[0003] In data analysis technology, machine learning and deep learning have become key technologies for improving prediction accuracy and automating data processing. Deep learning, especially its neural network model, is very suitable for processing and analyzing large amounts of unstructured data. The development of these technologies provides strong technical support for the intelligence and automation of CRM systems.

[0004] Existing CRM systems rely on structured data such as customer purchase history and demographic data to analyze customer behavior and market trends when processing product development data. However, this processing method still has technical problems such as insufficient accuracy. Summary of the invention

[0005] The present application provides a method, device, terminal and medium for processing production and sales product data, which are used to solve the technical problem that the existing CRM system has insufficient accuracy in the production and sales product development data processing method.

[0006] In order to solve the above technical problems, the first aspect of the present application provides a method for processing product data, comprising:

[0007] Collect structured data and unstructured data related to the target products and sales, wherein the structured data includes: customer purchase history records and customer attribute data, and the unstructured data includes: voice data, text data, and video data posted by customers;

[0008] By using a preset vector autoregression model, the structured data is processed according to the time series characteristics of the structured data to obtain production and sales product development forecast data;

[0009] By using a preset language macro model, language features are extracted and analyzed for the unstructured data to obtain production and marketing environment factor data of the target production and marketing product;

[0010] According to the production and sales product development forecast data and the production and sales environmental factor data, combined with a preset comprehensive forecast calculation formula, the development data forecast result of the target production and sales product is obtained.

[0011] Preferably, it also includes:

[0012] The structured data and the unstructured data are preprocessed.

[0013] Preferably, the preprocessing includes: missing value filling, outlier processing and data normalization.

[0014] Preferably, the comprehensive prediction calculation formula is specifically:

[0015]

[0016] In the formula, For the prediction results of the development data, is the weight coefficient, The product development forecast data output by the vector autoregression model, is the production and marketing environment factor data output by the language model, and b is the number of fine-tuning times of the language model.

[0017] Preferably, it also includes:

[0018] Calculating the Kalman gain vector of the vector autoregression model according to the collected structured data using a recursive least squares method;

[0019] According to the Kalman gain vector, in combination with a preset model update equation, the parameter vector and the covariance matrix of the vector autoregression model are updated.

[0020] Preferably, it also includes:

[0021] The large language model is fine-tuned and optimized through the LoRA algorithm.

[0022] Preferably, the vector autoregression model includes: a VAR model, a FAVAR model or a BVAR model.

[0023] At the same time, the second aspect of the present application provides a product data processing device, including:

[0024] A data collection unit, used to collect structured data and unstructured data related to the target product, wherein the structured data includes: customer purchase history records, customer attribute data, and the unstructured data includes: voice data, text data, and video data published by the customer;

[0025] A structured data processing unit, used to process the structured data according to the time series characteristics of the structured data through a preset vector autoregression model to obtain production and sales product development forecast data;

[0026] An unstructured data processing unit, used to extract and analyze language features of the unstructured data through a preset language macro model, so as to obtain production and marketing environment factor data of the target production and marketing product;

[0027] The comprehensive forecast processing unit is used to obtain the development data forecast result of the target production and sales product based on the production and sales product development forecast data and the production and sales environment factor data in combination with a preset comprehensive forecast calculation formula.

[0028] A third aspect of the present application provides a product data processing terminal, including: a memory and a processor;

[0029] The memory is used to store program codes, and the program codes correspond to a method for processing product data provided in the first aspect of the present application;

[0030] The processor is configured to read and execute the program code.

[0031] The fourth aspect of the present application provides a computer-readable storage medium, in which program code is stored. The program code is used to be read and executed by a processor to implement a production and sales product data processing method as provided in the first aspect of the present application.

[0032] It can be seen from the above technical solutions that this application has the following advantages:

[0033] The technical solution provided by this application first collects structured and unstructured data of target products and sales, and then combines the vector autoregression model and the large language model to process the structured data using the vector autoregression model, capture the dynamic relationship of multivariate time series data, and predict the future development of products and sales. Use the large language processing model to process unstructured data, extract deep-level features for prediction, and quantify external factors that affect the development of products and sales, such as customer preferences and market trends. Finally, the prediction results of the two models are combined to achieve the fusion of structured and unstructured data, and then dynamically adjust the weights to achieve a more accurate prediction of the development of products and sales. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 A flowchart of an embodiment of a method for processing product data provided in this application.

[0036] Figure 2 This is an overall logic block diagram of a method for processing production and sales product data provided in this application.

[0037] Figure 3 A logic block diagram for fine-tuning a large language model in a method for processing product data provided in this application.

[0038] Figure 4 A schematic structural diagram of an embodiment of a production and sales product data processing device provided in this application.

[0039] Figure 5 A schematic diagram of the structure of an embodiment of a production and sales product data processing terminal provided in this application. DETAILED DESCRIPTION

[0040] In the existing technology, the existing CRM system focuses on structured data analysis, but lacks the ability to analyze unstructured data. However, in the actual operation process, the types and quantity of customer data have also shown explosive growth, and customer-related unstructured data such as text, pictures, videos, etc. often contain rich information about customer preferences, market trends and potential risks. If not fully utilized, it will directly affect the accuracy and timeliness of the prediction.

[0041] In view of this, the present application provides a method, device, terminal and medium for processing production and sales product data, which are used to solve the technical problem of insufficient accuracy in the existing CRM system in processing production and sales product development volume data.

[0042] In order to make the purpose, features, and advantages of the invention of this application more obvious and easy to understand, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the embodiments described below are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0043] First, a detailed description of an embodiment of a method for processing product data provided by the present application is as follows:

[0044] See also Figure 1 and Figure 2 , a method for processing product data provided by an embodiment of the present application includes:

[0045] Step 101: Collect structured data and unstructured data related to target products and sales;

[0046] It should be noted that the technical solution provided in this embodiment can be implemented based on the customer relationship management system CRM, wherein the structured data includes: customer purchase history, customer attribute data, structured data: this part of the data covers multiple aspects of the customer's purchase history, including customer ID, purchase date, product ID, purchase quantity, purchase amount, etc. These data provide basic information about the customer's purchase behavior. At the same time, demographic data such as customer age, gender, income level, geographic location, etc. are also taken into consideration to gain a more comprehensive understanding of the customer's background. Inventory data includes product ID, inventory quantity, storage date, outbound date, etc. These data are crucial for predicting the development volume of production and sales products. In addition, sales data such as product ID, sales quantity, sales amount, sales date, etc. are also an indispensable part.

[0047] Unstructured data: Unstructured data includes voice data, text data, and video data posted by customers, such as customer telephone feedback (voice), product reviews (text), social media comments (video), etc. They provide in-depth information about customer preferences, market trends, and potential risks, and play an important role in improving the accuracy and timeliness of forecasts.

[0048] Step 102: Process the structured data according to the temporal characteristics of the structured data through a preset vector autoregression model to obtain production and sales product development forecast data;

[0049] The vector autoregression model is a statistical model that is particularly suitable for multivariate time series data. By jointly modeling multiple time series variables, the vector autoregression model is able to capture the dynamic relationship between variables, thereby more accurately predicting future trends. Use the formula of the vector autoregression model to model multivariate time series data to predict future production and sales development:

[0050]

[0051] in, is the variable vector at time t output by the vector autoregression model, is the coefficient matrix, is the hysteresis coefficient, is the error term, and c is a constant vector.

[0052] More specifically, the vector autoregression model mentioned in this embodiment preferably adopts a VAR model, but in some embodiments, a FAVAR model or a BVAR model may also be adopted.

[0053] Step 103: extract and analyze language features of the unstructured data using a preset language macro model to obtain production and marketing environment factor data of the target product;

[0054] It should be noted that in order to make full use of the rich information in unstructured data, open source large models (such as the LLaMA series) are selected for processing. At the very beginning of the model, open source pre-trained weights are used for deployment to quickly build a model with basic predictive capabilities. Then, the large model is regularly fine-tuned using newly collected unstructured data to better adapt it to the current market environment and customer needs. In the prediction process, the formula of the large model is used to model the unstructured data, and the features are extracted for prediction. Among them, the large model prediction structure is modeled using the following formula:

[0055]

[0056] in, Represent the text, video, and voice data of the product s, respectively. Represents a large model, Represents the parameters of the large model.

[0057] Step 104: Based on the product development forecast data and the product environment factor data, combined with a preset comprehensive forecast calculation formula, the development data forecast result of the target product is obtained.

[0058] It should be noted that the prediction results of structured data are finally integrated with the prediction results of unstructured data. Subsequently, a comprehensive prediction formula was designed to combine the two prediction results according to certain weights to obtain the final prediction results. Among them, the weight α is a dynamically adjusted parameter, which gradually increases with the increase of the number of fine-tuning b of the large model, but its upper limit is 0.5. To ensure that in the comprehensive prediction, both the prediction results of structured data and the prediction results of unstructured data occupy an important position, and the contribution of unstructured data will gradually increase with the number of fine-tuning, but the contribution of unstructured data will not be too high, causing the contribution of structured data to decrease. The expression of the final prediction result is as follows:

[0059]

[0060] In the formula, To develop data prediction results, is the weight coefficient, The output of the vector autoregression model is the forecast data of product development. Production and sales environment factor data output for the language model

[0061] in, , b is the number of times the large model is fine-tuned. As the number of fine-tuning increases, the contribution of the large model to the comprehensive prediction is gradually improved. This function can be adjusted according to actual use and fine-tuning frequency.

[0062] In some embodiments, in order to further improve the effect of the technical solution of this embodiment, the technical solution provided by this embodiment may further include the following technical contents:

[0063] After step 101, the method further includes:

[0064] Step 1010: pre-processing structured data and unstructured data;

[0065] In the preprocessing stage, the data are subjected to operations such as missing value filling, outlier processing, and data normalization to ensure the integrity and consistency of the data.

[0066] Furthermore, the method provided in this embodiment may further include:

[0067] According to the collected structured data, the Kalman gain vector of the vector autoregression model is calculated according to the recursive least square method;

[0068] According to the Kalman gain vector, combined with the preset model update equation, the parameter vector and covariance matrix of the vector autoregression model are updated.

[0069] It should be noted that in order to maintain the forecast accuracy and real-time performance of the VAR model, it is fine-tuned based on the latest structured data (such as daily sales and inventory data). The parameters of the VAR model are updated using recursive least squares (RLS) or other online learning algorithms to ensure that the model can adapt to the latest market environment and customer needs.

[0070] Specifically include: Calculate the Kalman gain vector:

[0071]

[0072] Update the parameter vector:

[0073]

[0074] Update the covariance matrix:

[0075]

[0076] in is the Kalman gain vector, which determines the impact of new observation data on the update of model parameters. When the Kalman gain is large, the impact of new data is large; otherwise, the impact is small. is the forgetting factor, which controls the importance of old data. When it is close to 1, the weight of old data is larger and the model is updated slowly; when it is smaller, the weight of new data is larger and the model is updated quickly. is the covariance matrix, which represents the uncertainty of the model parameters. As new data are added, it is gradually updated to reflect the accuracy of the current parameter estimation.

[0077] like Figure 3 As shown in the figure, for large models, the text data needs to be cleaned and processed first, including removing useless information such as stop words and punctuation marks, and performing text segmentation, word embedding and other operations to extract text features. For non-text data such as images and videos, feature extraction and processing are required. Then, the LoRA (Low-Rank Adaptation) algorithm is used to fine-tune the large model. The LoRA algorithm fine-tunes the model by adjusting the weights of the model by updating a low-rank matrix. During the training process, the cross entropy loss function is used to evaluate the predictive performance of the model, and the Adam optimizer is used to update and optimize the parameters of the model:

[0078]

[0079] in, is the original model weight, is a low-rank matrix. During the training process, the cross entropy loss function and the Adam optimizer are used to update the model parameters. is decomposed into the multiplication of two matrices ,in The dimension is , and the dimensions of A and B are and ,in , thus greatly improving the efficiency of fine-tuning.

[0080] Improve the model's online learning capabilities, automatically process and integrate new data, and perform real-time model updates and forecasts. This means that the system can adapt to the ever-changing market environment and customer needs, and provide accurate, real-time forecast results. At the same time, because the method has strong adaptability, it can be quickly deployed and applied in various CRM systems without specific environment adaptation. This makes the method widely applicable and practical, and can provide effective product development volume forecasting solutions for different industries and enterprises.

[0081] This application provides an innovative method for predicting the development volume of production and sales products in a customer relationship management system, which combines the VAR model and the large language model, especially for the comprehensive analysis and utilization of unstructured and structured data. This solution effectively combines the deep learning technology of artificial intelligence with traditional data processing technology, significantly improving the accuracy of predictions, data utilization efficiency, and cost-effectiveness of model training. In addition, through the fusion and fine-tuning of the model, this application enhances the flexibility of the system and its business decision-making support capabilities, enabling it to adapt to rapidly changing market conditions and complex data environments. Specific advantages include:

[0082] 1. Improve prediction accuracy: This application combines the VAR model and the large language model to make full use of unstructured and structured data to achieve higher prediction accuracy. In-depth analysis of unstructured data helps grasp customer preferences and market trends, while structured data provides quantitative business insights. The combination of the two significantly improves the overall prediction performance.

[0083] 2. High data utilization efficiency: By effectively integrating and analyzing multiple data types from different sources, this application significantly improves data utilization. This includes not only traditional structured data such as sales records and demographic data, but also unstructured data such as text reviews and social media content, thereby providing more comprehensive support for decision-making.

[0084] 3. Reduce training costs: Use large models for fine-tuning to avoid the high cost of training complex models from scratch. In addition, fix the parameters of the original pre-trained layers and only train the newly added layers, which not only shortens the training time but also reduces the consumption of computing resources.

[0085] 4. Enhanced business decision support: By integrating forecast results directly into the CRM system, this application provides enterprises with real-time and data-driven decision support. This helps optimize inventory management, adjust production plans, and develop more effective marketing strategies, thereby improving the company's response speed and market competitiveness.

[0086] 5. Flexibility and scalability: The method of this application is not only applicable to CRM systems, but can also be applied to forecasting needs in other fields, such as supply chain management, financial market analysis, etc. This cross-industry applicability provides a broader market application prospect.

[0087] The above is a detailed description of an embodiment of a method for processing product data provided by the present application. The following is a detailed description of an embodiment of a device for processing product data provided by the present application.

[0088] See also Figure 4 , a product data processing device provided in an embodiment of the present application includes:

[0089] The data collection unit 201 is used to collect structured data and unstructured data related to the target product, wherein the structured data includes: customer purchase history records and customer attribute data, and the unstructured data includes: voice data, text data and video data published by the customer;

[0090] The structured data processing unit 202 is used to process the structured data according to the time series characteristics of the structured data through a preset vector autoregression model to obtain the production and sales product development forecast data;

[0091] The unstructured data processing unit 203 is used to extract and analyze the language features of the unstructured data through a preset language model to obtain the production and marketing environment factor data of the target production and marketing products;

[0092] The comprehensive forecast processing unit 204 is used to obtain the development data forecast result of the target production and sales product based on the production and sales product development forecast data and the production and sales environment factor data in combination with a preset comprehensive forecast calculation formula.

[0093] In addition, the present application also provides detailed descriptions of embodiments such as a product and sales data processing terminal and a computer-readable storage medium related to the above-mentioned product and sales data processing method.

[0094] like Figure 5 As shown, an embodiment of the present application provides a product data processing terminal, the implementation types of the terminal include but are not limited to: a personal computer, an industrial computer, a server and an embedded intelligent device, and the main components of the terminal include: a memory 33 and a processor 31, wherein the memory 33 and the processor 31 can be connected via a communication bus 34;

[0095] The memory 33 is used to store program codes, which correspond to a method for processing product data in the above-mentioned embodiment;

[0096] The processor 31 is used to read and execute program codes.

[0097] The fourth aspect of the present application provides a computer-readable storage medium, in which program code is stored. The program code is used to be read and executed by a processor to implement a production and sales product data processing method as provided in the above embodiment.

[0098] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the terminals, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0099] In the several embodiments provided in the present application, it should be understood that the disclosed terminals, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0100] The terms "first", "second", "third", "fourth", etc. (if any) in the specification of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can, for example, be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0101] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.

[0102] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0103] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0104] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0105] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application 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. However, 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 embodiments of the present application.

Claims

1. A method for processing product data, characterized in that: include: Collect structured data and unstructured data related to the target products and sales, wherein the structured data includes: customer purchase history records and customer attribute data, and the unstructured data includes: voice data, text data, and video data posted by customers; By using a preset vector autoregression model, the structured data is processed according to the time series characteristics of the structured data to obtain production and sales product development forecast data; By using a preset language macro model, language features are extracted and analyzed for the unstructured data to obtain production and marketing environment factor data of the target production and marketing product; According to the production and sales product development forecast data and the production and sales environmental factor data, combined with a preset comprehensive forecast calculation formula, the development data forecast result of the target production and sales product is obtained.

2. A method for processing product data according to claim 1, characterized in that: Also includes: The structured data and the unstructured data are preprocessed.

3. A method for processing product data according to claim 2, characterized in that: The preprocessing includes: missing value filling, outlier processing and data normalization.

4. A method for processing product data according to claim 1, characterized in that: The comprehensive prediction calculation formula is specifically: In the formula, For the prediction results of the development data, is the weight coefficient, The product development forecast data output by the vector autoregression model, is the production and marketing environment factor data output by the language model, and b is the number of fine-tuning times of the language model.

5. A method for processing product data according to claim 1, characterized in that: Also includes: Calculating the Kalman gain vector of the vector autoregression model according to the collected structured data using a recursive least squares method; According to the Kalman gain vector, in combination with a preset model update equation, the parameter vector and the covariance matrix of the vector autoregression model are updated.

6. A method for processing product data according to claim 1, characterized in that: Also includes: The large language model is fine-tuned and optimized through the LoRA algorithm.

7. A method for processing product data according to claim 1, characterized in that: The vector autoregression model includes: a VAR model, a FAVAR model or a BVAR model.

8. A product data processing device, characterized in that: include: A data collection unit, used to collect structured data and unstructured data related to the target product, wherein the structured data includes: customer purchase history records, customer attribute data, and the unstructured data includes: voice data, text data, and video data published by the customer; A structured data processing unit, used to process the structured data according to the time series characteristics of the structured data through a preset vector autoregression model to obtain production and sales product development forecast data; An unstructured data processing unit, used to extract and analyze language features of the unstructured data through a preset language macro model, so as to obtain production and marketing environment factor data of the target production and marketing product; The comprehensive forecast processing unit is used to obtain the development data forecast result of the target production and sales product based on the production and sales product development forecast data and the production and sales environment factor data in combination with a preset comprehensive forecast calculation formula.

9. A product data processing terminal, characterized in that: include: Memory and processor; The memory is used to store program codes, and the program codes correspond to a method for processing product data according to any one of claims 1 to 7; The processor is configured to read and execute the program code.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, and the program code is used to be read and executed by a processor to implement a production and sales product data processing method as described in any one of claims 1 to 7.

Images