CN-121981747-A - Client intelligent marketing method and device, electronic equipment and storage medium

CN121981747ACN 121981747 ACN121981747 ACN 121981747ACN-121981747-A

Abstract

The application discloses a client intelligent marketing method, a device, electronic equipment and a storage medium, which relate to the technical field of digital marketing and comprise the steps of collecting and standardizing behavior event streams of enterprise clients on an electronic channel in real time, fusing static attributes, historical transaction and real-time behavior sequence data of the enterprise clients to construct multidimensional client characteristic representation, carrying out joint prediction through a pre-trained machine learning model to obtain a demand probability value aiming at a target marketing product, generating marketing decision instructions according to the probability value and business rules, and sending the marketing decision instructions to a current operation interface of the clients in real time for scene display. Through multi-source data fusion and real-time calculation, the problems of incomplete customer portraits, inaccurate demand prediction and missed marketing occasions caused by single dimension of data, static model and delayed marketing response are solved, and the technical effects of accurately predicting customer demands, immediately grasping marketing occasions and naturally integrating business scenes to improve user experience and conversion efficiency are achieved.

Inventors

QIAO GUANFENG
WU JING
GUO QING
ZHAO BOCHAO
WANG JINGYUAN

Assignees

中国建设银行股份有限公司河北省分行

Dates

Publication Date: 20260505
Application Date: 20251222

Claims (10)

1. A method for intelligent marketing to a customer, comprising: collecting behavior event streams generated by enterprise clients on electronic channels in real time, and carrying out standardized processing on the behavior event streams to generate standardized behavior events comprising client identifications, event types and occurrence time; based on the standardized behavior event, static attribute data, historical transaction data and real-time behavior sequence data of the enterprise client are fused, and a dynamically updated multidimensional client feature representation is constructed; Inputting the multi-dimensional customer characteristic representation into a pre-trained demand prediction model for joint prediction so as to acquire a demand probability value of the enterprise customer aiming at a target marketing product; And generating a marketing decision instruction according to the demand probability value and a preset business rule, and sending the marketing decision instruction to an electronic channel interface currently used by the enterprise client in real time for scene display.
2. The method for intelligent marketing to a customer according to claim 1, wherein the steps of collecting a real-time behavioral event stream generated by a customer of an enterprise on an electronic channel, and normalizing the behavioral event stream to generate a normalized behavioral event comprising a customer identification, an event type and an occurrence time comprise: asynchronously receiving client side behavior logs from a plurality of electronic channels through a message middleware; analyzing the behavior log, extracting client identifications, session identifications, event types, event time and page function identifications, and mapping the event types to a predefined standardized event set.
3. The method of claim 1, wherein the fusing static attribute data, historical transaction data, and real-time behavioral sequence data of the enterprise client based on the standardized behavioral events to construct a dynamically updated multidimensional client characteristic representation comprises: Partitioning the behavior event stream based on the client identification, and aggregating the continuously-occurring behavior events into a session sequence according to a preset session timeout time; Performing pattern matching on the behavior events in a sliding time window, and identifying a key event sequence conforming to a predefined behavior path; And splicing the key event sequence with the static attribute characteristics and the historical transaction statistical characteristics of the enterprise clients to form the multidimensional client characteristic representation.
4. The method of claim 1, wherein inputting the multi-dimensional customer feature representation into a pre-trained demand prediction model for joint prediction to obtain a demand probability value for the enterprise customer for a target marketing product comprises: The demand prediction model comprises a first prediction sub-model for processing structural features and a second prediction sub-model for processing behavior sequence features, wherein the first prediction sub-model is an integrated learning model based on a gradient lifting tree and is used for carrying out nonlinear modeling on numerical type and category type structural features, and the second prediction sub-model is a sequence learning model based on a cyclic neural network and is used for carrying out time sequence dependent modeling on a behavior event sequence; And fusing the outputs of the first predictor model and the second predictor model to obtain the demand probability value.
5. The intelligent marketing method according to claim 1, wherein the generating marketing decision instruction according to the demand probability value and the preset business rule comprises: combining and judging the demand probability value with the client value classification and the current behavior scene, and matching with a pre-configured marketing strategy rule; generating a marketing decision instruction containing recommended content, display positions and trigger conditions according to the matched marketing strategy rule; and the marketing decision instruction is issued to the electronic channel interface currently used by the enterprise client in real time for scene display.
6. The client intelligent marketing method in accordance with claim 1, further comprising: Collecting interactive feedback data of enterprise clients on marketing contents displayed in a scene; and updating a training sample set of the demand prediction model according to the interactive feedback data, and performing iterative optimization on the model to form a marketing effect closed-loop feedback mechanism.
7. A customer intelligent marketing apparatus, comprising: The system comprises an acquisition module, a data processing module and a data processing module, wherein the acquisition module is configured to acquire a behavior event stream generated by an enterprise client on an electronic channel in real time, and perform standardized processing on the behavior event stream to generate a standardized behavior event comprising a client identifier, an event type and an occurrence time; the construction module is configured to integrate static attribute data, historical transaction data and real-time behavior sequence data of the enterprise clients based on the standardized behavior events and construct a dynamically updated multidimensional client feature representation; the acquisition module is configured to input the multi-dimensional customer characteristic representation into a pre-trained demand prediction model to perform joint prediction so as to acquire a demand probability value of the enterprise customer for a target marketing product; and the issuing module is configured to generate a marketing decision instruction according to the demand probability value and a preset business rule, and issue the marketing decision instruction to an electronic channel interface currently used by the enterprise client in real time for scene display.
8. An electronic device, comprising: At least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the client smart marketing method of any of claims 1-6.
9. A non-transitory computer readable storage medium storing computer instructions for causing the computer to perform the customer intelligent marketing method in accordance with any one of claims 1-6.
10. A computer program product comprising a computer program which, when executed by a processor, implements the client smart marketing method according to any one of claims 1-6.

Description

Client intelligent marketing method and device, electronic equipment and storage medium Technical Field The application relates to the technical field of digital marketing, in particular to a client intelligent marketing method, a client intelligent marketing device, electronic equipment and a storage medium. Background With the acceleration of the transformation of the core channel of the banking service enterprise to the online electronic channel, the platforms such as the enterprise internet banking, the mobile banking and the like generate massive, real-time and multidimensional customer behavior data in the service process, and the data have higher timeliness and insight value compared with the traditional static enterprise information. However, the marketing management mode of the enterprise client is still rough in the existing bank, and the conventional bank mainly depends on personal experience of the client manager and few core financial indexes to judge, so that the requirement insight of large-scale and accurate is difficult to realize. Meanwhile, the traditional marketing system is mostly based on offline batch analysis, and cannot respond to real-time behaviors of clients in electronic channels such as foreign exchange quotation inquiry and frequent loan page browsing, so that marketing time is delayed and conversion rate is low. In addition, the existing customer portraits often integrate only static attribute and transaction result data, and lack effective fusion and dynamic modeling of process type data such as a customer behavior sequence in an electronic channel, so that the customer portraits are one-sided and updated slowly, and personalized recommendation of thousands of people and thousands of sides is difficult to support. More importantly, the marketing action is often disjointed with the flow of the actual use of the electronic channel by the client, and is often pushed through an independent channel, so that the marketing experience is hard and invasive, and the natural integration of service, namely marketing, cannot be realized. Disclosure of Invention The application provides a client intelligent marketing method, a client intelligent marketing device, electronic equipment and a storage medium. The method can solve the problems of incomplete customer portrait, inaccurate demand prediction, missed marketing opportunity and poor user experience caused by single data dimension, static model, lack of real-time property and unsound combination of marketing and electronic channels in the related technology. According to a first aspect of the present application, there is provided a client intelligent marketing method, comprising: collecting behavior event streams generated by enterprise clients on electronic channels in real time, and carrying out standardized processing on the behavior event streams to generate standardized behavior events comprising client identifications, event types and occurrence time; based on the standardized behavior event, static attribute data, historical transaction data and real-time behavior sequence data of the enterprise client are fused, and a dynamically updated multidimensional client feature representation is constructed; Inputting the multi-dimensional customer characteristic representation into a pre-trained demand prediction model for joint prediction so as to acquire a demand probability value of the enterprise customer aiming at a target marketing product; And generating a marketing decision instruction according to the demand probability value and a preset business rule, and sending the marketing decision instruction to an electronic channel interface currently used by the enterprise client in real time for scene display. According to a second aspect of the present application, there is provided a customer intelligent marketing apparatus comprising: The system comprises an acquisition module, a data processing module and a data processing module, wherein the acquisition module is configured to acquire a behavior event stream generated by an enterprise client on an electronic channel in real time, and perform standardized processing on the behavior event stream to generate a standardized behavior event comprising a client identifier, an event type and an occurrence time; the construction module is configured to integrate static attribute data, historical transaction data and real-time behavior sequence data of the enterprise clients based on the standardized behavior events and construct a dynamically updated multidimensional client feature representation; the acquisition module is configured to input the multi-dimensional customer characteristic representation into a pre-trained demand prediction model to perform joint prediction so as to acquire a demand probability value of the enterprise customer for a target marketing product; and the issuing module is configured to generate a marketing decision instruction according to the demand probabi