TWM601397U - Customized marketing system with customer clustering service - Google Patents

Abstract

A customized marketing system for customer grouping services. A processing module uses a deep learning algorithm to establish a pre-training model based on multiple digital trajectory training data stored in a storage module, and uses the pre-training model according to the Wait for the digital trajectory training data, generate multiple vector data, and then use a machine learning algorithm to establish a prediction model based on the vector data and the multiple transaction training data stored in the storage module, and use the prediction model according to Receive multiple digital trajectories to be analyzed from a server via a communication module, generate multiple prediction results, and obtain multiple target digital trajectories to be analyzed from the digital trajectory data to be analyzed based on the prediction results Finally, based on the digital trajectory data to be analyzed, a one-line sales list is generated.

Description

Customer segmentation service customized marketing system

This model relates to a marketing system, in particular to a customized marketing system for customer grouping services.

The existing financial industry conducts regular screening from the customer data in the database to find potential customers to establish a marketing list. Usually, the marketing department can quickly understand customer needs and then group potential customers.

However, the existing financial industry often uses fixed rules to filter out the list of potential customers and group them into groups. For example, customers who are between 25 and 40 years old and have purchased foreign currency funds will be determined by the system to purchase foreign currency funds later, so the system determines The customer is a potential customer and is classified as a group with high foreign exchange needs. However, the customer has purchased foreign currency funds in the past, which does not mean that they will purchase foreign currency funds in the future. The existing marketing screening methods cannot accurately identify potential customers .

Therefore, the purpose of this new model is to provide a method that can accurately identify potential Customer segmentation service customized marketing system for customers.

Therefore, the new customer grouping service customized marketing system is suitable for grouping multiple pieces of to-be-analyzed digital trajectory data stored in a server that are related to multiple customers to be grouped. Each piece of digital trajectory data to be analyzed includes multiple related The browsing record to be analyzed on a website includes a communication module, a storage module, and a processing module.

The communication module is connected to the server via a communication network.

The storage module stores multiple digital trajectory training data related to multiple customers, multiple transaction training data corresponding to the digital trajectory training data, and multiple digital trajectory data to be analyzed related to multiple customers to be grouped , Each digital trajectory training data includes multiple browsing records related to the website, and each transaction training data includes multiple commodity purchase records.

The processing module is electrically connected to the communication module and the storage module. The processing module uses a deep learning algorithm to create a pre-training model for converting the data into vectors based on the digital trajectory training data, and Using the pre-training model, based on the digital trajectory training data, multiple vector data corresponding to the digital trajectory training data are generated, and then a machine learning algorithm is used to establish the transaction training data and the vector data. A prediction model for predicting whether a customer will purchase a product, and using the prediction model to generate a plurality of prediction results corresponding to the waiting analysis digital trajectory data based on the waiting analysis digital trajectory data, and according to the prediction results, In the digital trajectory data waiting to be analyzed, multiple target digital trajectory data to be analyzed are obtained, The prediction result corresponding to the digital trajectory data of the targets to be analyzed is that goods will be purchased, and finally a one-line sales list is generated based on the digital trajectory data of the targets to be analyzed.

The effect of the present invention is that the processing module accurately finds out the digital trajectory data to be analyzed for the targets that will purchase the goods according to the prediction results.

1: Customer segmentation service customized marketing system

11: Communication module

12: Storage module

13: Processing module

21~27: Steps

261~263: Step

The other features and effects of the present invention will be clearly presented in the implementation with reference to the drawings, in which: Figure 1 is a block diagram illustrating an embodiment of the present new customer segmentation service customized marketing system; Figure 2 is A flowchart illustrating the flow of steps performed by this embodiment of the new type of customer grouping service customized marketing system; and Fig. 3 is a flowchart assisting Fig. 2 to illustrate the sub-steps of step 26.

Referring to Figure 1, an embodiment of the new customer grouping service customized marketing system 1 is suitable for grouping multiple pieces of digital trajectory data to be analyzed related to multiple customers to be grouped stored in a server 2. The analyzed digital trajectory data includes a plurality of browsing records to be analyzed related to a website. The customer segmentation service customized marketing system 1 includes a communication module 11, a storage module 12, and a processing module 13. Worth noting In this embodiment, the customer grouping service customized marketing system 1 is, for example, a computer host, but it is not limited to this.

The communication module 11 is connected to the server 2 via a communication network 100. The communication network 100 is, for example, an Internet or an intranet.

It is worth noting that in this embodiment, the server 2 sets up a website with multiple pages, and embeds a tracking code (Tracking Code) on the website to collect the waiting-to-analyze digital trajectory data, but not Is limited.

The storage module 12 stores multiple digital trajectory training data respectively related to multiple customers, multiple transaction training data corresponding to the digital trajectory training data, and multiple digital trajectories to be analyzed related to multiple customers to be grouped. Data, each digital trajectory training data includes multiple browsing records related to the website, and each transaction training data includes multiple commodity purchase records. It is worth noting that in this embodiment, each browsing record includes a browsing page related to one of the pages of the website, a session ID, a client ID, and a message related to the browsing device. Records (cookies), a page view time, and a previous page view. Each product purchase record includes a customer identification code, a transaction type, a transaction amount, a handling fee, and a transaction time, but not Is limited.

The processing module 13 is electrically connected to the communication module 11 and the storage module 12.

1 and 2 to illustrate the flow of steps executed by this embodiment of the new customer segmentation service customized marketing system 1.

In step 21, the processing module 13 uses a deep learning algorithm to establish a pre-training model for converting the data into vectors based on the digital trajectory training data. It is worth noting that, in this embodiment, the deep learning algorithm is, for example, an embedded language model (Embeddings from Language Models, ELMo) deep learning algorithm related to Natural Language Processing (NLP), but Not limited to this.

In step 22, the processing module 13 uses the pre-training model to generate a plurality of vector data corresponding to the digital trajectory training data according to the digital trajectory training data. It is worth noting that, in this embodiment, each vector data includes a plurality of word vectors related to the browsing records of the corresponding digital trajectory training data, but it is not limited to this.

In step 23, the processing module 13 uses a machine learning algorithm such as tree-based models (Tree-Based Models) and random forest (Random Forest) algorithms to establish based on the transaction training data and the vector data A prediction model used to predict whether a customer will buy a product.

In step 24, the processing module 13 uses the prediction model to generate a plurality of predictions respectively corresponding to the waiting digital trajectory data received from the server 2 via the communication module 11 As a result, each prediction result is that the product will be purchased or the product will not be purchased.

In step 25, the processing module 13 determines from the prediction results from the In the to-be-analyzed digital trajectory data, a plurality of target to-be-analyzed digital trajectory data is obtained, and the predicted result corresponding to the to-be-analyzed digital trajectory data of the target is that a commodity will be purchased.

In step 26, the processing module 13 uses a grouping algorithm to group the target digital trajectory data to be analyzed to obtain multiple groups.

3 in conjunction, step 26 includes sub-steps 261 to 263. The sub-steps of step 26 are described below.

In step 261, the processing module 13 uses a sequence alignment (Sequence Alignment) algorithm to group the target digital trajectory data to be analyzed to obtain multiple trajectory groups.

In step 262, for each trajectory group, the processing module 13 uses the pre-training model to generate multiple target digits to be analyzed corresponding to the trajectory group according to the target to be analyzed digital trajectory data of the trajectory group. Target vector data of the trajectory data.

In step 263, the processing module 13 uses the grouping algorithm to group the trajectory groups according to the target vector data corresponding to the trajectory groups to obtain the groups.

It is worth noting that, in this embodiment, the processing module 13 first uses a gene sequence alignment algorithm to subdivide the cluster in step 261, and uses the clustering algorithm (such as the k-means algorithm) in step 263. Rough grouping. For example, there are 3578 target digital trajectory data to be analyzed. After subdividing the group in step 261, the processing module 13 Groups the 3578 target digital trajectory data to be analyzed into 326 trajectory groups, and after rough grouping in step 263, the processing module 13 groups the 326 trajectory groups into 7 groups.

In step 27, the processing module 13 generates a sales list based on the target to-be-analyzed digital trajectory data and the groups.

In summary, the new customer segmentation service customized marketing system uses the processing module 13 to establish the pre-training model based on the digital trajectory training data, and use the pre-training model to generate the corresponding digital trajectory training data Then use the prediction model to generate the prediction results based on the real-time waiting-analyzed digital trajectory data received by the communication module 11 from the server 2, and finally find out accurately based on the prediction results Will purchase the target’s to-be-analyzed digital trajectory data, and use the target’s to-be-analyzed digital trajectory data to generate the marketing list, so that the marketing department can generate real-time customized push services based on the marketing list, so it can indeed To achieve a new purpose.

However, the above-mentioned are only examples of the present model, and should not be used to limit the scope of implementation of the present model, all simple equivalent changes and modifications made in accordance with the patent scope of the present model application and the contents of the patent specification still belong to This new patent covers the scope.

1: Customer segmentation service customized marketing system

11: Communication module

12: Storage module

13: Processing module

Claims (5)

A customer segmentation service customized marketing system, which is suitable for grouping multiple pieces of to-be-analyzed digital track data stored in a server that are related to multiple customers to be grouped. Each piece of digital-track data to be analyzed includes multiple pieces of data related to a website Browsing records to be analyzed include: a communication module connected to the server via a communication network; a storage module that stores multiple training data of digital trajectories related to multiple clients, and multiple records corresponding to the digital trajectories The transaction training data of the training data, and multiple pieces of to-be-analyzed digital trajectory data related to multiple customers to be grouped. Each digital trajectory training data includes multiple browsing records related to the website, and each transaction training data includes multiple products Purchase records; and a processing module, which is electrically connected to the communication module and the storage module, and uses a deep learning algorithm to establish a pre-training model for converting the data into vectors based on the digital trajectory training data, And use the pre-training model to generate a plurality of vector data corresponding to the digital trajectory training data based on the digital trajectory training data, and then use a machine learning algorithm based on the transaction training data and the vector data, Establish a predictive model for predicting whether a customer will purchase a product, and use the predictive model to generate a plurality of digital trajectories corresponding to the waiting analysis digital trace data received from the server via the communication module According to the prediction results of the data, a plurality of target digital trajectory data to be analyzed are obtained from the digital trajectory data waiting to be analyzed. The corresponding prediction result is that the product will be purchased, and finally a sales list is generated based on the digital trajectory data to be analyzed for these targets. The customer segmentation service customized marketing system according to claim 1, wherein the deep learning algorithm is a deep learning algorithm for embedded language models related to natural language processing. The customer grouping service customized marketing system according to claim 1, wherein the website includes a plurality of pages, each browsing record includes a browsing page related to one of the pages of the website, and each vector data includes a plurality of The word vector of the browsing page related to the browsing record of the corresponding digital trajectory training data. The customer grouping service customized marketing system according to claim 1, wherein the processing module uses a grouping algorithm to group the target digital trajectory data to be analyzed to obtain multiple groups, and the processing The module also generates the marketing list based on these groups. The customer grouping service customized marketing system according to claim 4, wherein the processing module uses a genetic sequence comparison algorithm to group the target digital trajectory data to be analyzed to obtain multiple trajectory groups, For each trajectory group, the processing module uses the pre-training model and generates a plurality of vector data corresponding to the target digital trajectory data to be analyzed in the trajectory group according to the target digital trajectory data of the trajectory group , And then use the grouping algorithm to group the trajectory groups according to the vector data corresponding to the trajectory groups to obtain the groups.

Images