TWM596917U - Co-marketing recommendation system - Google Patents

Abstract

A cooperative marketing recommendation system includes a transaction server and a marketing integration device. The transaction server receives the transfer information of a co-sale customer from the co-sale client and sends it to the marketing integration device. The marketing integration device determines whether a collection account of the co-marketing customer transfer information matches a customer account of multiple customer data, and after determining that there is a match, utilizes a customer community based on the matched customer data The network grouping model generates a customer grouping result, and uses a product preference classification model to generate a product preference classification result, and then determines at least one target from multiple pieces of marketing activity information based on the customer grouping result and the product preference classification result Marketing activity information.

Description

Co-marketing recommendation system

This new type is about a recommendation system, especially a cooperative marketing recommendation system.

In the past, when financial institutions carried out product marketing, they could only promote existing customers through specific channels. In order to make the promotion objects not limited to existing customers, financial institutions used the concept of co-marketing system to make marketing activities unnecessary. Customer marketing can achieve the effect of diffusion.

The concept of the co-marketing system is that a co-marketing customer transmits the marketing activity information of a financial institution. If the customer is interested in the marketing activity information, the purpose of indirect promotion is achieved.

However, the effectiveness of marketing campaign promotion is often limited by the degree of customer demand for the product, and now the co-marketing customer does not have an effective way to obtain the customer's preference, and then provide marketing activity information that meets the customer's needs.

Therefore, the purpose of the new model is to provide a collaborative marketing recommendation system that can effectively obtain customer preferences.

Therefore, the new collaborative marketing recommendation system is connected to a collaborative marketing client via a first communication network. The collaborative marketing recommendation system includes a transaction server and a marketing integration device.

The transaction server connects to the co-marketing client via the first communication network, and receives a co-marketing client transfer information from the co-marketing client, the co-marketing client transfer information includes a co-marketing client account number, a transfer amount, and A collection account.

The marketing integration device stores multiple pieces of customer data corresponding to multiple customers, and connects to the transaction server via a second communication network. Each customer data includes a customer account number, at least one transfer information, and at least one marketing activity history participation information , At least one product purchase record information, and multiple pieces of marketing activity information, such customer data includes a co-sale customer data related to the co-sale client, after receiving the co-sale customer transfer information from the transaction server, The marketing integration device determines whether there is a match between the collection account of the co-sale customer transfer information and the customer accounts of the customer data, and is determining that the collection account and the customer account of the customer data exist After a match, based on the matching customer data transfer information and marketing activity historical participation information, a customer social network grouping model for dividing customers into multiple different groups is generated to generate a customer grouping result and use a product The preference classification model generates a product preference classification result. The product preference classification includes multiple product categories and multiple probabilities corresponding to the product categories, and then based on the customer grouping results and the product preference classification results, from these marketing activities At least one target marketing activity information is determined from the information and sent to the transaction server, so that the transaction server sends the at least one target marketing activity information to the co-marketing client.

The function of the present invention lies in that the marketing integration device determines the at least one target marketing that meets the needs of customers from the marketing activity information by using the customer social network grouping model and the product preference classification model according to the matched customer data Activity Information.

Referring to FIG. 1, an embodiment of the new cooperative marketing recommendation system includes a transaction server 11, a line marketing integration device 12, and a modeling device 13.

The transaction server 11 is connected to a co-marketing client 101 via a first communication network 100. It is worth noting that the first communication network 100 is, for example, the Internet, but it is not limited thereto.

The marketing integration device 12 is connected to the transaction server 11 via the first communication network 100, and stores multiple pieces of customer data corresponding to multiple customers, each customer data includes a customer account number, at least one transfer information, and at least one marketing activity Historical participation information, at least one product purchase record information, and multiple pieces of marketing activity information. The customer data includes a co-marketing customer data related to the co-marketing client 101. Each product purchase record information includes a purchased product category, a purchase amount, and a product balance, where the product balance refers to the product balance of a specific product at a certain point in time, and a certain point in time can be defined as the end of each month or Other continuous time, but not limited to this.

The modeling device 13 is connected to the marketing integration device 12 via a second communication network 102, and stores multiple pieces of customer social network group model training data and multiple pieces of product preference classification model training data, and each customer social network group model training The data includes training transfer information and at least one training marketing activity information. Each product preference classification model training data includes multiple training scores corresponding to multiple product categories and one of these product categories. It is worth noting that the second communication network 102 is, for example, an enterprise intranet, and the purchase product category included in each product purchase record information is one of these product categories, such as foreign exchange survival, Product categories such as fixed foreign exchange deposits, funds, insurance, gold passbooks, credit cards, and credits, but not limited to this. It should be further noted that in other embodiments, each product preference classification model training data also includes a customer basic data, a channel behavior, and a customer group label, where the customer basic data includes, for example, age, gender, occupation, Place of residence, and annual income, the channel behavior is, for example, the amount and frequency of bank branch use, or the amount and frequency of bank ATM use, and the customer group label is, for example, payroll transfer notes, online banking use notes, or active account notes, But not limited to this.

Referring to FIG. 1 and FIG. 2, a step flow of a modeling program executed by this embodiment of the new cooperative marketing recommendation system will be described.

In step 21, the modeling device 13 uses an unsupervised machine learning algorithm to generate a customer social network group for dividing customers into multiple different groups based on the training data of the customer social network group models Model, and send the customer social network grouping model to the marketing integration device 12. It is worth noting that the unsupervised machine learning algorithm includes, for example, Principal component analysis (Principal component analysis) or cluster analysis (Clustering), but not limited to this.

In step 22, the modeling device 13 uses a supervised machine learning algorithm to generate a product preference classification model based on the product preference classification model training data, and transmits the product preference classification model to the marketing integration device 12 . It is worth noting that the supervised machine learning algorithm includes, for example, a random forest algorithm, but not limited to this.

It should be particularly noted that, in this embodiment, step 21 is before step 22, but in other embodiments, step 22 may be performed before step 21 or simultaneously, and is not limited to this.

Referring to Figs. 1 and 3, a step flow of a co-marketing program executed by this embodiment of the new co-marketing recommendation system will be described.

In step 31, the affiliate client 101 transmits an affiliate client transfer information to the transaction server 11, the affiliate client transfer information is related to the affiliate client 101 transfer to a user (not shown) Transfer information, including a co-sale customer account, a transfer amount, and a collection account.

In step 32, the transaction server 11 transmits the co-sale customer transfer information to the marketing integration device 12.

In step 33, the marketing integration device 12 determines whether there is a match between the collection account number of the co-marketing customer transfer information and the customer account numbers of the customer data. When the marketing integration device 12 determines that there is a match between the collection account number of the co-marketing customer transfer information and the customer accounts of the customer data, the process proceeds to step 34; and when the marketing integration device 12 determines the cooperation If there is no match between the payment account number of the customer transfer information and the customer account information of the customer information, the process proceeds to step 36.

In step 34, the marketing integration device 12 uses the customer social network grouping model to generate a customer grouping result based on the matching customer data transfer information and marketing activity history participation information.

In step 35, the marketing integration device 12 generates a product preference classification result using the product preference classification model based on the product purchase record information of the matched customer data. The product preference classification includes the product categories and multiple Should wait for the probability of the product category.

Referring to FIG. 4 together, step 35 further includes sub-steps 351 and 352. The sub-steps of step 35 are described below.

In step 351, the marketing integration device 12 obtains a plurality of category scores corresponding to the product categories according to the product purchase record information of the matched customer data. It is worth noting that, in this embodiment, the category scores are obtained by a preset model. The preset model gives the category scores according to the weight of each field of similar customers, but not limited to this.

In step 352, the marketing integration device 12 uses the product preference classification model according to the category scores to generate the product preference classification result.

It should be particularly noted that, in other embodiments, each customer profile also includes a customer basic profile, a channel behavior, and a customer group tag. The marketing integration device 12 also uses the customer basic profile of the matched customer profile, The channel behavior, and the customer group label, use the product preference classification model to generate the product preference classification result.

In step 36, the marketing integration device 12 generates a customer grouping result using the customer social network grouping model based on the transfer information of the co-marketing customer data and the history participation information of the marketing activity.

In step 37, the marketing integration device 12 generates a product preference classification result using the product preference classification model based on the product purchase record information of the co-marketing customer data. It is worth noting that steps 36 and 37 indicate that if the marketing integration device 12 does not store customer data corresponding to the payment account number, it will be replaced with the co-marketing customer data according to the principle of similar groups.

Referring to FIG. 5 together, step 37 further includes sub-steps 371 and 372. The sub-steps of step 37 are described below.

In step 371, the marketing integration device 12 obtains a plurality of category scores corresponding to the product categories based on the product purchase record information of the co-marketing customer data. It is worth noting that, in this embodiment, similar to step 351, the category scores are obtained by the default model. The default model gives the category scores based on the weights of similar customer fields. Limited.

In step 372, the marketing integration device 12 uses the product preference classification model according to the category scores to generate the product preference classification result.

It should be particularly noted that, in other embodiments, the co-marketing customer data also includes a co-marketing customer basic data, a co-marketing channel behavior, and a co-marketing customer group label, and the marketing integration device 12 is also based on the co-marketing The basic information of the co-sales customer, the behavior of the co-sales channel, and the label of the co-sales customer group of the customer data, using the product preference classification model, to generate the product preference classification result.

It should be further noted that in this embodiment, step 34 is before step 35, step 36 is before step 37, in other embodiments, step 34 can be performed after step 35 or at the same time, step 36 can be after step 37 Or at the same time, not limited to this.

In step 38 after step 35 or step 37, the marketing integration device 12 determines at least one target marketing activity information from the marketing activity information based on the customer grouping result and the product preference classification result, and combines the at least one The target marketing activity information is sent to the transaction server 11.

In step 39, the transaction server 11 transmits the at least one target marketing activity information to the co-marketing client 101, so that the co-marketing client 101 transmits a transfer account and the at least one target marketing activity information to the user terminal .

It is worth noting that the user terminal can be directed to a marketing webpage based on the at least one target marketing activity information, and the marketing integration device 12 will receive a feedback message from a web server related to the marketing webpage. The integration device 12 creates customer data related to the user terminal according to the feedback message.

In summary, the new collaborative marketing recommendation system uses the marketing integration device 12 to generate the customer by using the customer social network clustering model and the product preference classification model based on the matched customer data or the collaborative customer data The grouping result and the product preference classification result, and based on the customer grouping result and the product preference classification result, the at least one target marketing activity information that meets the customer's needs is determined from the marketing activity information, so it is indeed possible to achieve a new type of cost purpose.

However, the above are only examples of the new model. When the scope of the new model cannot be limited by this, any simple equivalent changes and modifications made according to the patent application scope and patent specification content of the new model are still regarded as Within the scope of this new patent.

1: Co-marketing recommendation system 11: Transaction server 12: Marketing integration device 13: Modeling device 100: the first communication network 101: Co-marketing client 102: Second communication network 21, 22: steps 31~39: Step 351, 352: steps 371, 372: steps

Other features and functions of the present invention will be clearly presented in the embodiments with reference to the drawings, in which: FIG. 1 is a block diagram illustrating an embodiment of the new cooperative marketing recommendation system; FIG. 2 is a flowchart illustrating how to implement a modeling procedure of this embodiment of the new co-marketing recommendation system; FIG. 3 is a flowchart illustrating how to implement a co-marketing program of this embodiment of the new co-marketing recommendation system; 4 is a flowchart to assist in explaining the sub-steps of step 35 of FIG. 3; and FIG. 5 is a flowchart to assist in explaining the sub-steps of step 37 of FIG. 3.

1: Co-marketing recommendation system

11: Transaction server

12: Marketing integration device

13: Modeling device

100: the first communication network

101: Co-marketing client

102: Second communication network

Claims (5)

A cooperative marketing recommendation system is connected to a cooperative marketing client via a first communication network. The system includes: A transaction server, connected to the co-branding client via the first communication network, receiving a co-branding client transfer information from the co-branding client, the co-branding client transfer information including a co-branding client account number, a transfer amount, and A collection account; A marketing integration device that stores multiple customer data corresponding to multiple customers and connects to the transaction server via a second communication network. Each customer data includes a customer account number, at least one transfer information, and at least one marketing activity history participation Information, at least one product purchase record information, and multiple pieces of marketing activity information, such customer information includes a co-sale customer data related to the co-sale client, after receiving the co-sale customer transfer information from the transaction server , Determine whether the collection account number of the co-sale customer transfer information has a match with the customer accounts of the customer data, and after determining that there is a match between the collection account number and the customer accounts of the customer data , Based on the transfer information of the matching customer data and the historical participation information of marketing activities, using a customer social network grouping model for dividing customers into multiple different groups, generating a customer grouping result, and using a product preference classification model , A product preference classification result is generated, the product preference classification includes multiple product categories and multiple probabilities corresponding to the product categories, and then determined from the marketing activity information based on the customer grouping result and the product preference classification result At least one target marketing activity information is output and transmitted to the transaction server, so that the transaction server transmits the at least one target marketing activity information to the co-marketing client. The co-marketing recommendation system as described in claim 1, wherein the marketing integration device determines that there is no match between the collection account number of the co-marketing customer transfer information and the client accounts of the client data, according to the The transfer information of the co-marketing customer data and the historical participation information of marketing activities, use the customer social network grouping model to generate a customer grouping result, and according to the product purchase record information of the co-marketing customer data, use a product preference classification model to generate A product preference classification result. The product preference classification includes multiple product categories and multiple probabilities corresponding to the product categories, respectively. As described in claim 1 or 2, wherein the co-marketing recommendation system described in claim 1, wherein the marketing integration device obtains a plurality of products corresponding to the product purchase record information based on the matched customer data The category score of the product category, and then based on the category score, use the product preference classification model to generate the product preference classification result. The co-marketing recommendation system as described in claim 1, further includes: A modeling device that stores multiple pieces of customer social network grouping model training data, and is connected to the marketing integration device via the second communication network. Each customer social network grouping model training data includes a training transfer information and at least one training marketing Event information, based on the training data of these customer social network cluster models, uses an unsupervised machine learning algorithm to generate the customer social network cluster model, and transmits the customer social network cluster model to the marketing integration device. The co-marketing recommendation system as described in claim 1, further includes: A modeling device that stores multiple pieces of product preference classification model training data and is connected to the marketing integration device via the second communication network. Each product preference classification model training data includes a plurality of training scores corresponding to the product categories and the One of the product categories, etc., based on the training data of the product preference classification model, uses a supervised machine learning algorithm to generate the product preference classification model, and transmits the product preference classification model to the marketing integration device.

Images