Data modeling method and system based on intelligent marketing scene

Abstract

The invention discloses a data modeling method and system based on an intelligent marketing scene, and belongs to the technical field of data processing, and the method comprises the steps: firstly determining the marketing conversion probability and behavior habit information of a user through integrating marketing data and user data; then, a marketing prediction network is constructed, and the network is trained and evaluated by using the information so as to acquire network performance parameters. And when the network performance satisfies a preset constraint, storing the network performance as a marketing prediction model. The model can accept marketing scheme input and carry out marketing effect prediction. Compared with the prior art, the method has the advantages that the marketing precision and the credibility of effect prediction are remarkably improved, and meanwhile, the marketing decision-making efficiency is improved. By providing personalized products and services, the method also helps to enhance user experience. Besides, the intelligent development of the marketing field is promoted, the positive influence on the improvement of the competitiveness of the whole industry is achieved, and powerful support is provided for an enterprise to formulate an efficient and accurate marketing strategy in the intelligent marketing era.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and more specifically to a data modeling method and system based on intelligent marketing scenarios. Background Art

[0002] With the rapid development of science and technology, especially the widespread application of big data and artificial intelligence technology, intelligent marketing has become an important means for enterprises to improve their market competitiveness and marketing efficiency. Intelligent marketing collects and analyzes consumer behavior data, consumption habits, scenario information, etc., to build accurate user portraits, and achieve personalized recommendations and precision marketing. As one of the core technologies of intelligent marketing, data modeling plays an important role in improving marketing effectiveness and optimizing customer experience.

[0003] At present, research in the field of intelligent marketing has made significant progress, especially in the construction of user portraits, content recommendation systems and marketing effect prediction. By processing and analyzing massive user behavior data, such as browsing records and purchase records, accurate user portraits are constructed, providing strong support for precision marketing in e-commerce and other fields. At the same time, the content recommendation system based on user portraits and interest preferences generates recommendation models through machine learning algorithms such as collaborative filtering and deep learning, which improves user satisfaction and corporate conversion rates. However, intelligent marketing still faces some challenges, including data quality issues, such as inaccurate and incomplete data and difficulty in data integration; weak model generalization ability, difficult to adapt to new data or different scenarios; and high technical barriers, involving complex technology and high cost investment, which poses a challenge to small and medium-sized enterprises.

[0004] Therefore, how to provide a data modeling method and system that can integrate multi-source heterogeneous data is an urgent problem that needs to be solved by technical personnel in this field. Summary of the invention

[0005] In view of this, the present invention provides a data modeling method and system based on intelligent marketing scenarios, which optimize the intelligent marketing decision-making process through data modeling and intelligent prediction technology, and improve marketing efficiency and effectiveness.

[0006] In order to achieve the above object, the present invention adopts the following technical solution:

[0007] In one aspect, the present invention provides a data modeling method based on an intelligent marketing scenario, comprising:

[0008] Obtain marketing data and user data;

[0009] Determine a marketing conversion probability of the user based on the marketing data and the user data;

[0010] Determining the user's behavior habit information based on the user data;

[0011] Constructing a marketing prediction network, training and evaluating the marketing prediction network with the marketing conversion probability and the behavior habit information, and obtaining network performance parameters;

[0012] If the network performance parameters satisfy the preset performance parameter constraints, storing the marketing prediction network as a marketing prediction model;

[0013] The marketing plan is input into the marketing prediction model to predict the marketing effect.

[0014] Preferably, determining the marketing conversion probability of the user based on the marketing data and the user data includes:

[0015] Merging the marketing data and the user data by user ID to generate a comprehensive user data set;

[0016] Extracting the user's transaction data, the user's interaction data with marketing activities, and marketing product data based on the user comprehensive data set;

[0017] Determine the user's purchase data and effective marketing data based on the interaction data, the transaction data and the marketing commodity data;

[0018] The marketing conversion probability of the user is determined according to the purchase data, the effective marketing data and the marketing product data.

[0019] Preferably, determining the user's purchase data and effective marketing data according to the interaction data, the transaction data and the marketing commodity data includes:

[0020] Performing feature processing on the interaction data, the transaction data and the marketing commodity data respectively to obtain an interaction commodity feature matrix, a purchase commodity feature matrix and a marketing commodity feature matrix;

[0021] The interactive product feature matrix and the purchased product feature matrix are concatenated to obtain a purchased feature matrix, and the interactive product feature matrix and the marketing product feature matrix are concatenated to obtain an interested feature matrix;

[0022] Purchase data is obtained according to the purchase feature matrix, and effective marketing data is obtained according to the interest feature matrix.

[0023] Preferably, constructing a marketing prediction network includes a prediction output layer and a deep network input layer and a shallow network input layer simultaneously connected to the prediction output layer.

[0024] Preferably, the marketing prediction network includes at least one of a Deep Crossing architecture network model and a Wide & Deep architecture network model.

[0025] Preferably, after predicting the marketing effect based on the marketing prediction model, it also includes generating marketing suggestions based on the marketing prediction results and the behavioral habit information.

[0026] On the other hand, the present invention provides a data modeling system based on an intelligent marketing scenario, which is used to implement the above-mentioned data modeling method based on an intelligent marketing scenario, including:

[0027] Data collection module, used to obtain marketing data and user data;

[0028] A marketing conversion probability module, used to determine the marketing conversion probability of the user based on the marketing data and the user data;

[0029] A behavior habit extraction module, used to determine the behavior habit information of the user based on the user data;

[0030] A prediction network construction module is used to construct a marketing prediction network, train and evaluate the marketing prediction network with the marketing conversion probability and the behavior habit information, and obtain network performance parameters;

[0031] A storage module, configured to store the marketing prediction network as a marketing prediction model if the network performance parameter satisfies a preset performance parameter constraint;

[0032] The prediction module is used to input the marketing plan into the marketing prediction model to predict the marketing effect.

[0033] Preferably, the marketing conversion probability module includes:

[0034] A data merging unit, used to merge the marketing data and the user data through the user ID to generate a user comprehensive data set;

[0035] A comprehensive data extraction unit, configured to extract the transaction data of the user, the interaction data between the user and the marketing activities, and the marketing commodity data based on the user comprehensive data set;

[0036] An effective data extraction unit, configured to determine the user's purchase data and effective marketing data based on the interaction data, the transaction data and the marketing commodity data;

[0037] A probability conversion calculation unit is used to determine the user's marketing conversion probability based on the purchase data, the effective marketing data and the marketing product data.

[0038] Preferably, the system further comprises a suggestion module for generating marketing suggestions based on the marketing prediction results and the behavioral habit information.

[0039] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a data modeling method and system based on intelligent marketing scenarios, which accurately determines the user's marketing conversion probability and behavioral habit information by integrating marketing data and user data, and then builds and optimizes the marketing prediction network model. This method not only improves the accuracy of marketing and the credibility of effect prediction, but also improves the efficiency of marketing decision-making, and helps to provide personalized products and services to enhance user experience. In addition, it promotes the intelligent development of the marketing field and contributes to the improvement of the competitiveness of the entire industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention 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 embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0041] Figure 1 A flow chart of a data modeling method based on an intelligent marketing scenario provided by the present invention.

[0042] Figure 2 A framework diagram of a data modeling system based on an intelligent marketing scenario provided by the present invention. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] The embodiment of the present invention discloses a data modeling method based on intelligent marketing scenarios. Figure 1 As shown, including:

[0045] Obtain marketing data and user data; marketing data and user data are collected and integrated from multiple channels to ensure the comprehensiveness and accuracy of the data; marketing data usually includes historical marketing activity data, advertising data, sales data, etc., which can reflect the company's past marketing efforts and results. User data covers the user's personal information (such as age, gender, region, etc.), purchase history, browsing history, user behavior logs, etc., which help understand the user's preferences and needs.

[0046] Determine the user's marketing conversion probability based on marketing data and user data; marketing conversion probability refers to the probability that a user will convert into actual purchase behavior and purchase intention after receiving specific marketing information. This step uses data analysis and mining technology, machine learning or statistical models to combine marketing data and user data to calculate the marketing conversion probability of each user. This probability value can reflect the user's sensitivity and reaction to different marketing information, and is an important input for the subsequent construction of a marketing prediction network.

[0047] Determine the user's behavior habit information based on user data; user behavior habit information refers to the consumption habits and behavior patterns that users show in their daily lives, including purchase channels, channels for obtaining product information, etc. This step analyzes user data to extract key information such as user preferences, activity, purchase cycle, purchase categories, etc., to form a comprehensive understanding of user behavior habits. This information helps companies grasp user needs more accurately and formulate personalized marketing strategies.

[0048] Build a marketing prediction network, train and evaluate the marketing prediction network with marketing conversion probability and behavior habit information, and obtain network performance parameters; in this step, it is necessary to use deep learning or machine learning technology to build a marketing prediction network. This network uses marketing conversion probability and behavior habit information as input features, and learns and predicts through a complex neural network structure. During the training process, the network is trained using historical data, and the network performance is evaluated through cross-validation, loss function optimization, and other means. Network performance parameters include accuracy, recall, F1 score, etc. These parameters can reflect the prediction ability and stability of the network.

[0049] If the network performance parameters meet the preset performance parameter constraints, the marketing prediction network is stored as a marketing prediction model;

[0050] Input the marketing plan into the marketing forecasting model to predict the marketing effect.

[0051] Specifically, determining the user's marketing conversion probability based on the marketing data and the user data includes:

[0052] Merge marketing data and user data through user ID to generate a comprehensive user data set; this step is to closely link marketing activities and user behavior.

[0053] Based on the user comprehensive data set, the user's transaction data, the user's interaction data with marketing activities, and the marketing product data are extracted; the extracted key information includes the user's transaction data, the user's interaction data with marketing activities, and the marketing product data. The transaction data records the user's purchase history, including the purchase time, purchased products, and purchase amount; the interaction data reflects the user's response to the marketing activities, such as whether the user clicked on the advertisement, browsed the promotion page, participated in the lucky draw, the number of product clicks, etc.; and the marketing product data provides the basic information of the products involved in the marketing, such as price, category, discount, inventory, etc.

[0054] Determine the user's purchase data and effective marketing data based on interaction data, transaction data and marketing product data; purchase data refers to the record of the user's actual purchase behavior, while effective marketing data refers to the record of marketing activities that successfully guide users to generate purchasing intentions or behaviors.

[0055] Determine the user's marketing conversion probability based on purchase data, effective marketing data, and marketing product data.

[0056] Furthermore, the user's purchase data and effective marketing data are determined based on the interaction data, transaction data and marketing product data, including:

[0057] Perform feature processing on the interaction data, transaction data and marketing product data to obtain an interaction product feature matrix, a purchase product feature matrix and a marketing product feature matrix;

[0058] The interactive product feature matrix and the purchased product feature matrix are concatenated to obtain the purchased feature matrix, and the interactive product feature matrix and the marketing product feature matrix are concatenated to obtain the interested feature matrix;

[0059] Purchase data is obtained based on the purchase feature matrix, and effective marketing data is obtained based on the interest feature matrix.

[0060] In another embodiment, constructing a marketing prediction network includes a prediction output layer and a deep network input layer and a shallow network input layer simultaneously connected to the prediction output layer.

[0061] Specifically, the marketing prediction network includes at least one of a Deep Crossing architecture network model and a Wide & Deep architecture network model.

[0062] The Deep Crossing architecture network model is known for its powerful deep learning capabilities. It can automatically extract high-level abstract features from raw data and capture complex associations between data through a multi-layer neural network structure. This model is particularly suitable for processing high-dimensional sparse data, such as user behavior logs and product attribute information, thereby showing excellent performance in predicting user marketing conversion probability. The Deep Crossing architecture network model may include a Stacking layer, a Multiple Residual Units layer, and a Scoring layer, wherein the Stacking layer is used to splice the input marketing features and user behavior features to form an input vector of the Multiple Residual Units layer, the Multiple ResidualUnits layer adopts a standard 18-layer residual network (ResNet 18) structure, and the Scoring layer adopts logistic regression to convert the prediction results into a probability value for output. At the same time, the loss function of the Deep Crossing architecture network model is a logarithmic loss function. Specifically, in one embodiment, the marketing features and user behavior features are both statistical information quantities, and therefore belong to dense features, so in this embodiment, the Deep Crossing architecture network model does not have an Embedding layer.

[0063] The Wide&Deep architecture network model combines the memory capacity of the linear model and the generalization capacity of the deep learning model. The Wide part is responsible for capturing the direct relationship between the user's historical behavior patterns and product attributes, and improving the accuracy of predictions by memorizing frequently occurring feature combinations; while the Deep part is responsible for learning potential and complex feature interactions from massive data, and mining new feature combinations through nonlinear transformations of deep neural networks. This combination enables the Wide&Deep model to have good generalization capabilities and adaptability to new data while maintaining high prediction accuracy.

[0064] When building a marketing prediction network, depending on the specific application scenario and data characteristics, you can choose to use the Deep Crossing architecture or the Wide & Deep architecture alone, or combine the two to take full advantage of their respective advantages. Such a design not only improves the prediction performance of the model, but also enhances the flexibility and scalability of the model, providing strong support for intelligent marketing decisions.

[0065] Furthermore, after predicting the marketing effect based on the marketing prediction model, it also includes generating marketing suggestions based on the marketing prediction results and behavioral habit information.

[0066] Behavioral information, including but not limited to users' purchase history, browsing preferences, interaction frequency, and response time, provides valuable insights into users' real needs and preferences. By deeply analyzing the association between these behavioral information and marketing forecast results, it is possible to intelligently identify which marketing strategies are most likely to attract specific user groups, which products or services are more popular in a specific time period, and how to maximize marketing effects by adjusting the content, form, and timing of marketing information.

[0067] Based on this comprehensive consideration, a series of customized marketing recommendations are generated. These recommendations may include recommending specific marketing channels (such as social media, email marketing, or offline events), optimizing the copywriting and visual design of marketing information, adjusting the pricing strategy of products or services, and developing personalized promotions. These recommendations are designed to not only improve the conversion rate of marketing activities, but also enhance user engagement and satisfaction, thereby building a more solid and lasting customer relationship.

[0068] In short, marketing recommendations generated based on marketing forecast results and behavioral habit information are the bridge between data analysis and actual actions. They make the company's marketing strategy more accurate, efficient and humane, laying a solid foundation for the company's market competitiveness and long-term development.

[0069] On the other hand, the present invention provides a data modeling system based on intelligent marketing scenarios, which is used to implement the above-mentioned data modeling method based on intelligent marketing scenarios, such as Figure 2 As shown, including:

[0070] Data collection module, used to obtain marketing data and user data;

[0071] A marketing conversion probability module, used to determine the marketing conversion probability of a user based on marketing data and user data;

[0072] A behavior habit extraction module is used to determine the user's behavior habit information based on the user data;

[0073] The prediction network construction module is used to build a marketing prediction network, train and evaluate the marketing prediction network with marketing conversion probability and behavior habit information, and obtain network performance parameters;

[0074] A storage module, used for storing the marketing prediction network as a marketing prediction model if the network performance parameters meet the preset performance parameter constraints;

[0075] The prediction module is used to input the marketing plan into the marketing prediction model to predict the marketing effect.

[0076] Preferably, the marketing conversion probability module includes:

[0077] A data merging unit, used to merge marketing data and user data through user IDs to generate a comprehensive user data set;

[0078] A comprehensive data extraction unit, used to extract user transaction data, user-marketing activity interaction data, and marketing product data based on the user comprehensive data set;

[0079] An effective data extraction unit, used to determine the user's purchase data and effective marketing data based on the interaction data, transaction data and marketing commodity data;

[0080] The probability conversion calculation unit is used to determine the user's marketing conversion probability based on purchase data, effective marketing data and marketing product data.

[0081] Preferably, the system also includes a suggestion module for generating marketing suggestions based on the marketing prediction results and behavioral habit information.

[0082] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0083] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A data modeling method based on intelligent marketing scenarios, characterized in that: include: Obtain marketing data and user data; Determine a marketing conversion probability of the user based on the marketing data and the user data; Determining the user's behavior habit information based on the user data; Constructing a marketing prediction network, training and evaluating the marketing prediction network with the marketing conversion probability and the behavior habit information, and obtaining network performance parameters; If the network performance parameters satisfy the preset performance parameter constraints, storing the marketing prediction network as a marketing prediction model; The marketing plan is input into the marketing prediction model to predict the marketing effect.

2. According to the data modeling method based on intelligent marketing scenarios according to claim 1, it is characterized in that: Determining a marketing conversion probability of the user based on the marketing data and the user data includes: Merging the marketing data and the user data by user ID to generate a comprehensive user data set; Extracting the user's transaction data, the user's interaction data with marketing activities, and marketing product data based on the user comprehensive data set; Determine the user's purchase data and effective marketing data based on the interaction data, the transaction data and the marketing commodity data; The marketing conversion probability of the user is determined according to the purchase data, the effective marketing data and the marketing product data.

3. The data modeling method based on intelligent marketing scenarios according to claim 2 is characterized in that: Determining the user's purchase data and effective marketing data according to the interaction data, the transaction data, and the marketing commodity data includes: Performing feature processing on the interaction data, the transaction data and the marketing commodity data respectively to obtain an interaction commodity feature matrix, a purchase commodity feature matrix and a marketing commodity feature matrix; The interactive product feature matrix and the purchased product feature matrix are concatenated to obtain a purchased feature matrix, and the interactive product feature matrix and the marketing product feature matrix are concatenated to obtain an interested feature matrix; Purchase data is obtained according to the purchase feature matrix, and effective marketing data is obtained according to the interest feature matrix.

4. The data modeling method based on intelligent marketing scenarios according to claim 1 is characterized in that: The marketing prediction network includes a prediction output layer and a deep network input layer and a shallow network input layer connected to the prediction output layer.

5. The data modeling method based on intelligent marketing scenarios according to claim 4 is characterized in that: The marketing prediction network includes at least one of a Deep Crossing architecture network model and a Wide & Deep architecture network model.

6. The data modeling method based on intelligent marketing scenarios according to claim 1 is characterized in that: After predicting the marketing effect based on the marketing prediction model, it also includes generating marketing suggestions based on the marketing prediction results and the behavioral habit information.

7. A data modeling system based on intelligent marketing scenarios, characterized in that: A data modeling method based on an intelligent marketing scenario for implementing any one of claims 1 to 6, comprising: Data collection module, used to obtain marketing data and user data; A marketing conversion probability module, used to determine the marketing conversion probability of the user based on the marketing data and the user data; A behavior habit extraction module, used to determine the behavior habit information of the user based on the user data; A prediction network construction module is used to construct a marketing prediction network, train and evaluate the marketing prediction network with the marketing conversion probability and the behavior habit information, and obtain network performance parameters; A storage module, configured to store the marketing prediction network as a marketing prediction model if the network performance parameter satisfies a preset performance parameter constraint; The prediction module is used to input the marketing plan into the marketing prediction model to predict the marketing effect.

8. The data modeling system based on intelligent marketing scenarios according to claim 7 is characterized in that: The marketing conversion probability module includes: A data merging unit, used to merge the marketing data and the user data through the user ID to generate a user comprehensive data set; A comprehensive data extraction unit, configured to extract the transaction data of the user, the interaction data between the user and the marketing activities, and the marketing commodity data based on the user comprehensive data set; An effective data extraction unit, configured to determine the user's purchase data and effective marketing data based on the interaction data, the transaction data and the marketing commodity data; A probability conversion calculation unit is used to determine the user's marketing conversion probability based on the purchase data, the effective marketing data and the marketing product data.

9. The data modeling system based on intelligent marketing scenarios according to claim 7, characterized in that: The system also includes a suggestion module for generating marketing suggestions based on the marketing prediction results and the behavioral habit information.

Images