Banking business marketing optimization method and system based on artificial intelligence

Abstract

The invention relates to a banking business marketing optimization method and system based on artificial intelligence, and the method comprises the steps: carrying out the intention recognition of user input, and obtaining an input type; analyzing the user input according to the input category to obtain an embedded vector input by the user; calculating the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and generating an optimal marketing scheme in combination with the semantic similarity and the operation effect quantitative index for generating the marketing scheme; calculating a current operation effect quantitative index according to a marketing effect after marketing of the optimal marketing scheme, inserting a marketing scheme meeting an access standard, and rearranging each scheme of the target knowledge base according to the operation effect quantitative index to obtain an optimized target knowledge base, after marketing is finished, a knowledge base optimization module is combined to rearrange the marketing schemes of the knowledge base after each marketing activity, marketing influence factors can be found more quickly by comparing marketing cases with similar operation effect quantitative indexes, and banking business marketing optimization is facilitated.

Description

Technical Field

[0001] This invention relates to the field of business marketing plan generation, specifically to a method and system for optimizing banking business marketing based on artificial intelligence. Background Technology

[0002] Retail banking involves numerous long-cycle marketing campaigns. While the initial development cycle for these campaigns is lengthy, the workflow remains largely consistent after stable deployment. Subsequent development focuses shift to comparing the effectiveness of different campaign versions, iterating marketing messages and recommendation strategies, and updating campaign plans accordingly. To achieve precise marketing, during the strategy formulation phase, business personnel, in collaboration with data analysts, conduct extensive customer segmentation, configuring strategies accordingly. During the strategy development phase, they design customized marketing messages, posters, and WeChat articles for each customer segment. This process presents a challenge in balancing the degree of customization with the timeliness of the marketing campaigns.

[0003] With the development of artificial intelligence, existing technologies use large language models to identify user needs. Based on the identification results, the intelligent agent generates corresponding marketing plans and provides them to bank staff using customer profile systems and solution data in the database. This AI + staff dialogue model allows staff to collaboratively develop marketing campaigns by interacting with the intelligent agent, mitigating the negative impact of increased time costs associated with higher customization levels. However, it lacks campaign review and makes it difficult to promptly apply the review results to the next marketing cycle.

[0004] Therefore, there is an urgent need for a marketing optimization method in the banking industry that can promptly conduct marketing reviews and iterate and update marketing activities in real time. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a method and system for optimizing banking business marketing based on artificial intelligence. By introducing quantitative indicators of business performance, the marketing effect of each marketing plan is quantified, and the marketing plans are reordered when updating the knowledge base. This allows high-quality marketing cases to be prioritized in similar situations, and by comparing marketing cases with similar quantitative indicators of business performance, the marketing influencing factors can be found more quickly, which is beneficial to the optimization of banking business marketing.

[0006] This invention is achieved through the following technical solution:

[0007] On the one hand, this invention provides an artificial intelligence-based method for optimizing banking business marketing, which includes:

[0008] S10: Perform intent recognition on user input to obtain the input category;

[0009] S20: Parse the user input according to the input category to obtain the embedding vector of the user input;

[0010] S30: Calculate the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and combine the semantic similarity with the quantitative indicators of the operational effectiveness of the generated marketing plan to generate the optimal marketing plan;

[0011] S40: Calculate the current business performance quantification index based on the marketing effect after the marketing of the optimal marketing plan, insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the business performance quantification index to obtain the optimized target knowledge base.

[0012] Furthermore, the generation of the optimal marketing plan based on highly relevant data includes:

[0013] The data are sorted according to semantic similarity, and the top-ranked data in the TOPK1 are selected as candidate data according to the preset TOPK1 to obtain a first candidate data group.

[0014] The first candidate data group is reranked to calculate the correlation between each candidate data and the user input. The candidate data ranked in the top 2 of the preset TOPK2 are selected as the optimal candidate data to obtain the optimal candidate data group.

[0015] The data groups are sorted according to the quantitative indicators of the best candidate data groups, and the top 3 candidate data groups with the highest quantitative indicators of the best candidate data groups are used as the final data groups.

[0016] The optimal marketing plan is obtained by processing the optimal data set using a large language model.

[0017] Further, step S40 includes:

[0018] S401.A: Calculate the improvement in marketing success rate based on the success rate of the sub-customer groups in the optimal marketing plan and the success rate of the sub-customer groups in the control group:

[0019] S401.B: Calculate the improvement rate of customer management objectives based on the customer marketing target data of the optimal marketing plan, the customer marketing target data of the control group, and the importance of each customer marketing objective;

[0020] S402: Calculate quantitative indicators of business performance based on the improvement in marketing success rate, the improvement in customer management goals, and customer grouping;

[0021] S403: Insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the quantitative indicators of business performance to obtain the optimized target knowledge base.

[0022] Further, in step S401.A, the improvement in the average marketing success rate T of the customer group is calculated using the following formula:

[0023]

[0024] Where n is the total number of sub-customer groups, C n S represents the number of people in each sub-group. i To optimize the success rate of marketing to sub-customer groups in the marketing strategy, The marketing success rate for the sub-customer group is the control group.

[0025] Furthermore, in step S401.B, the improvement rate R of the customer group management target is calculated using the following formula:

[0026]

[0027] Among them, O i Customer marketing target data for the optimal marketing plan K represents the customer marketing target data for the control group. i The importance of targeting specific customer groups and marketing objectives.

[0028] Further, in step S402, the quantitative indicator W of operating performance is calculated using the following formula:

[0029]

[0030] Among them, C n Let T represent the number of people in each sub-customer group, T represent the improvement in the average marketing success rate of the customer group, and R represent the improvement in the customer group's business objectives.

[0031] On the other hand, the present invention also provides an artificial intelligence-based banking business marketing optimization system, characterized in that it includes:

[0032] Intent recognition module: used to recognize the intent of user input and obtain the input category;

[0033] Text parsing module: used to parse user input according to the input category and obtain the embedding vector of the user input;

[0034] Marketing plan generation module: used to calculate the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and to generate the optimal marketing plan by combining the semantic similarity and the quantitative indicators of the business performance of the generated marketing plan;

[0035] Knowledge base optimization module: This module is used to calculate the current business performance quantification index based on the marketing effect after the marketing of the optimal marketing plan, insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the business performance quantification index to obtain the optimized target knowledge base.

[0036] Furthermore, the knowledge base optimization module includes:

[0037] Marketing Success Rate Improvement Calculation Submodule: Used to calculate the marketing success rate improvement based on the marketing success rate of the sub-customer group of the optimal marketing plan and the marketing success rate of the sub-customer group of the control group.

[0038] Customer segment management target improvement calculation submodule: used to calculate the improvement of customer segment management target based on the customer segment marketing target data of the optimal marketing plan, the customer segment marketing target data of the control group, and the importance of each customer segment marketing target;

[0039] The Business Performance Quantification Submodule is used to calculate quantitative indicators of business performance based on the improvement in marketing success rate, the improvement in customer management goals, and customer grouping.

[0040] The knowledge base optimization submodule is used to insert marketing plans that meet the admission criteria and rearrange each plan in the target knowledge base according to the quantitative indicators of business performance to obtain the optimized target knowledge base.

[0041] On the other hand, this application also provides a computer-readable storage medium, wherein when the computer program is executed by a processor, it implements the steps of an artificial intelligence-based banking marketing optimization method as described in any of the above claims.

[0042] This invention proposes an AI-based banking marketing optimization system and method. The AI-based system introduces quantitative indicators of operational effectiveness, quantifying the marketing effect of each marketing campaign. After semantic retrieval of user-input questions, the system re-ranks the quantitative indicators of operational effectiveness for each marketing campaign, prioritizing high-quality marketing cases in similar situations. A large language model is then used to further rank these high-quality cases, resulting in the marketing campaign best matched to the user's question, effectively improving the success rate of marketing campaigns. Furthermore, after the marketing campaign concludes, a knowledge base optimization module re-ranks the marketing campaigns based on the knowledge base after each marketing activity. This not only eliminates the need to traverse the knowledge base during campaign generation, optimizing campaign generation time, but also allows for faster identification of marketing influencing factors by comparing marketing cases with similar quantitative indicators of operational effectiveness, thus facilitating banking marketing optimization.

[0043] To better understand and implement this invention, the following detailed description is provided in conjunction with the accompanying drawings. Attached Figure Description

[0044] Figure 1 A structural block diagram of the AI-based banking marketing optimization system provided by this invention;

[0045] Figure 2 To execute Figure 1 The flowchart shown is a process for optimizing banking marketing based on artificial intelligence in the inspection system.

[0046] Figure 3 A flowchart for the script generation stage of the exemption from historical duplicate data provided by the present invention;

[0047] Figure 4 A structural block diagram of an artificial intelligence-based banking marketing optimization system provided in another embodiment of the present invention; Detailed Implementation

[0048] To address the lack of effective marketing optimization methods for banking operations in existing technologies, which enable timely marketing reviews, the inventors analyzed banking marketing systems. They discovered that current AI-based marketing plan generation technologies primarily rely on large language models to identify user questions. Based on the identified results, RAG (Retrieval Augmentation) technology performs a coarse screening of a large dataset in a knowledge base, yielding a batch of potentially relevant documents. These documents are then ranked based on simple similarity scores such as vector distance and keyword matching. However, this ranking, which only considers surface similarity, easily prioritizes irrelevant or low-quality documents, impacting the accuracy and efficiency of responses.

[0049] Furthermore, in customer segmentation scenarios, the correlation between segments is extremely strong. After initial screening, the scores of each data point remain very similar, with little differentiation. Setting the top K too high can cause token overflow, while setting it too low can lead to the rejection of high-quality marketing cases. Therefore, simply storing the results of marketing reviews into the corresponding knowledge base is insufficient to guarantee that high-quality marketing strategies can be retrieved, and it does not provide any positive benefit to marketing reviews.

[0050] To address the aforementioned issues, this invention provides an AI-based system and method for optimizing banking marketing. By introducing quantitative indicators of operational effectiveness, the system quantifies the marketing impact of each marketing campaign and reorders the campaigns when updating the knowledge base. This allows high-quality marketing cases to be prioritized in similar situations, and by comparing marketing cases with similar quantitative indicators of operational effectiveness, the system can more quickly identify marketing influencing factors, which is beneficial for optimizing banking marketing.

[0051] Specifically, please refer to Figure 1This AI-based banking marketing optimization system includes: an intent recognition module 10, a text parsing module 20, a marketing plan generation module 30, and a knowledge base optimization module 40. The intent recognition module identifies the intent of the user's input question, selects a suitable intelligent assistant and the corresponding knowledge base for targeted knowledge retrieval. The text parsing module processes the context of the user's input question, extracts feature information from the text, and converts this information into an embedded vector representation that is easy for computers to process. The marketing plan generation module 30 uses RAG technology to retrieve content from the knowledge base using the embedded vectors, and finally generates an accurate answer based on an LLM model. The knowledge base optimization module 40 updates the types of marketing plans in the knowledge base and ranks the marketing plans based on the effectiveness of the optimal plan selected for each marketing campaign, resulting in higher quality marketing plans generated for subsequent banking marketing activities. Please refer to [link / reference]. Figure 2 In a banking marketing campaign, the execution flow of each component of the banking marketing optimization system is as follows:

[0052] The intent recognition module 10 is used to perform step S10: to recognize the intent of the user input and obtain the input category;

[0053] User questions cover a variety of areas, such as dynamic customer segment analysis, marketing material generation, and customer management strategies. Different user questions will generate different answers. By categorizing common questions, we can improve the accuracy and efficiency of subsequent searches.

[0054] The text parsing module 20 is used to perform step S20: parse the user input according to the input category to obtain the embedding vector of the user input.

[0055] After obtaining the input category, the user can select the corresponding search AI assistant. For example, the dynamic customer group analysis AI assistant can complete customer group segmentation; the marketing material AI assistant can complete the preparation of marketing materials; the present invention takes the customer group management concept as an example, corresponding to the strategy library search AI assistant. The strategy library search AI assistant embeds the user's input question to obtain the corresponding embedding vector expression.

[0056] The marketing plan generation module 30 is used to perform step S30: calculate the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and generate the optimal marketing plan by combining the semantic similarity and the quantitative indicators of the business effect of the generated marketing plan.

[0057] RAG technology first calculates semantic similarity by combining the embedding vector with documents in the knowledge base connected to the strategy library retrieval AI assistant. Then, it sorts the data based on semantic similarity and selects the top-ranked data from a pre-defined TOPK1 as candidate data, resulting in a first candidate data group. Next, it performs reranking on this first candidate data group, calculating the relevance of each candidate data to the user input. Based on a pre-defined TOPK2, it selects the top-ranked candidate data from the second set as the optimal candidate data, resulting in the optimal candidate data group. Reranking involves concatenating the user question and each candidate data and cross-coding them in real time to calculate a relevance score. This allows for a deeper understanding and capture of the semantic relevance of the text. Subsequently, it sorts the optimal candidate data group according to the business performance quantification indicators, using the top three candidate data with the highest business performance quantification indicator values ​​as the final data, resulting in the final data group. Finally, it applies a large language model to the optimal data group to obtain the optimal marketing plan.

[0058] The calculation of semantic similarity and reranking are actually semantic relevance analysis, with the focus on filtering out irrelevant marketing plans. By using the quantitative indicators of business performance, several sets of plans with high matching degree are selected. This process can quickly find the sets of plans that have the best effect on solving user problems. The quantitative indicators of business performance pre-rank the high-quality marketing plans, prioritize the most effective marketing data plan, and then use a large language model to match the part of each plan that is most relevant to the user problem to obtain the optimal marketing plan.

[0059] The knowledge base optimization module 40 is used to execute step S40: calculate the current business performance quantitative index based on the marketing effect after the marketing of the optimal marketing plan, insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the business performance quantitative index to obtain the optimized target knowledge base.

[0060] Please see Figure 3 and Figure 4 The knowledge base optimization module 40 includes: 401.A Marketing Success Rate Improvement Calculation Submodule 401.A, which is used to execute step S401.A: Calculate the marketing success rate improvement T based on the marketing success rate of the sub-customer group of the optimal marketing plan and the marketing success rate of the sub-customer group of the control group.

[0061] In a marketing campaign, the known conditions are: the complete set of customer acquisition objectives U = {U1, U2, U3, ..., U...} N}, where U N Potential business metrics include: AUM asset growth rate, AUM asset achievement rate, asset holding growth rate, recurring product purchase rate, customer asset retention rate, mobile banking new signing rate, social security card activation rate, etc., constraining UN Dimensionless standardization has been performed, i.e., U N ∈[0,1]. Marketing objective O targeting a specific customer group. ′ ={O1,O2,O3,…,O M If M≤N, and the marketing objective O is for new customer acquisition, then... ′ ={AUM asset achievement rate, mobile banking new signing rate, and fixed-term product purchase rate}, based on the definition, the marketing objective O for the action group and the marketing objective for the control group can be calculated. Business objectives for a specific customer group ′ The importance of the corresponding marketing objective is K = {K1, K2, K3, ..., K} M}, M≤N,K i ≥K i+1 The control group represents marketing activities conducted without additional marketing efforts; the normal business operations of the control group reflect its marketing objectives.

[0062] After segmentation, the customer base is divided into n sub-groups, with the number of people in each sub-group being C1, C2, ..., Cn. n The marketing success rates of the action groups were S1, S2, ..., S... n The marketing success rate of the control group was The specific formula for the marketing success rate improvement T is as follows:

[0063]

[0064] Where n is the total number of sub-customer groups, C n S represents the number of people in each sub-group. i To optimize the success rate of marketing to sub-customer groups in the marketing strategy, The marketing success rate for the sub-customer group is the control group.

[0065] The customer segment management target improvement calculation submodule 401.B is used to execute step S401.B: Based on the customer segment marketing target data of the optimal marketing plan, the customer segment marketing target data of the control group, and the importance of each customer segment marketing target, calculate the customer segment management target improvement; calculate the customer segment management target improvement R using the following formula:

[0066]

[0067] Among them, O i Customer marketing target data for the optimal marketing plan K represents the customer marketing target data for the control group. i The importance of targeting specific customer groups and marketing objectives.

[0068] The submodule 402 for quantifying operational effectiveness is used to execute step S402: calculating the quantitative indicators of operational effectiveness based on the improvement in marketing success rate, the improvement in customer group management goals, and customer group segmentation. Considering that the sample size affects the reliability of the improvement in average customer group marketing success rate (T) and the improvement in customer group management goals (R), a weighting coefficient is added to the quantitative indicator W of operational effectiveness. The quantitative indicator W of operational performance is calculated using the following formula:

[0069]

[0070] Among them, C n Let T represent the number of people in each sub-customer group, T represent the improvement in the average marketing success rate of the customer group, and R represent the improvement in the customer group's business objectives.

[0071] The knowledge base optimization submodule 403 is used to execute step S403: to insert marketing plans that meet the admission criteria, and to rearrange each plan in the target knowledge base according to the quantitative indicators of business performance to obtain the optimized target knowledge base.

[0072] After rearrangement, marketing plans with similar marketing effects can be quickly identified. By comparing the differences between different marketing plans, the influencing factors affecting marketing effectiveness can be found, thereby assisting bank employees in flexibly adjusting marketing plans and improving marketing optimization efficiency.

[0073] This invention proposes an AI-based banking marketing optimization system and method. The AI-based system introduces quantitative indicators of operational effectiveness, quantifying the marketing effect of each marketing campaign. After semantic retrieval of user-input questions, the system re-ranks the quantitative indicators of operational effectiveness for each marketing campaign, prioritizing high-quality marketing cases in similar situations. A large language model is then used to further rank these high-quality cases, resulting in the marketing campaign best matched to the user's question, effectively improving the success rate of marketing campaigns. Furthermore, after the marketing campaign concludes, a knowledge base optimization module re-ranks the marketing campaigns based on the knowledge base after each marketing activity. This not only eliminates the need to traverse the knowledge base during campaign generation, optimizing campaign generation time, but also allows for faster identification of marketing influencing factors by comparing marketing cases with similar quantitative indicators of operational effectiveness, thus facilitating banking marketing optimization.

[0074] This invention can take the form of a computer program product implemented on one or more storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing program code. Computer-readable storage media include permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to: phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0075] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and the present invention also intends to include these modifications and variations.

Claims

1. A method for optimizing banking business marketing based on artificial intelligence, characterized in that, include: S10: Perform intent recognition on user input to obtain the input category; S20: Parse the user input according to the input category to obtain the embedding vector of the user input; S30: Calculate the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and combine the semantic similarity with the quantitative indicators of the operational effectiveness of the generated marketing plan to generate the optimal marketing plan; S40: Calculate the current business performance quantification index based on the marketing effect after the marketing of the optimal marketing plan, insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the business performance quantification index to obtain the optimized target knowledge base.

2. The AI-based banking marketing optimization method according to claim 1, characterized in that, The process of generating the optimal marketing plan based on highly relevant data includes: The data are sorted according to semantic similarity, and the top-ranked data in the TOPK1 are selected as candidate data according to the preset TOPK1 to obtain a first candidate data group. The first candidate data group is reranked to calculate the correlation between each candidate data and the user input. The candidate data ranked in the top 2 of the preset TOPK2 are selected as the optimal candidate data to obtain the optimal candidate data group. The data groups are sorted according to the quantitative indicators of the best candidate data groups, and the top 3 candidate data groups with the highest quantitative indicators of the best candidate data groups are used as the final data groups. The optimal marketing plan is obtained by processing the optimal data set using a large language model.

3. The AI-based banking marketing optimization method according to claim 2, characterized in that, Step S40 includes: S401.A: Calculate the improvement in marketing success rate based on the success rate of the sub-customer groups in the optimal marketing plan and the success rate of the sub-customer groups in the control group: S401.B: Calculate the improvement rate of customer management objectives based on the customer marketing target data of the optimal marketing plan, the customer marketing target data of the control group, and the importance of each customer marketing objective; S402: Calculate quantitative indicators of business performance based on the improvement in marketing success rate, the improvement in customer management goals, and customer grouping; S403: Used to insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the quantitative indicators of business performance to obtain an optimized target knowledge base.

4. The AI-based banking marketing optimization method according to claim 3, characterized in that: In step S401.A, the average marketing success rate improvement T of the customer group is calculated using the following formula: Where n is the total number of sub-customer groups, C n S represents the number of people in each sub-group. i To optimize the success rate of marketing to sub-customer groups in the marketing strategy, The marketing success rate for the sub-customer group is the control group.

5. The AI-based banking marketing optimization method according to claim 4, characterized in that, In step S401.B, the improvement rate R of customer group management target is calculated using the following formula: Wherein, represents the customer marketing target data for the optimal marketing plan, ---- represents the customer marketing target data for the control group, and represents the importance of the corresponding customer marketing target.

6. The AI-based banking marketing optimization method according to claim 5, characterized in that, In step S402, the quantitative indicator W of operating performance is calculated using the following formula: Among them, C n Let T represent the number of people in each sub-customer group, T represent the improvement in the average marketing success rate of the customer group, and R represent the improvement in the customer group's business objectives.

7. A banking business marketing optimization system based on artificial intelligence, characterized in that, include: Intent recognition module: used to recognize the intent of user input and obtain the input category; Text parsing module: used to parse user input according to the input category and obtain the embedding vector of the user input; Marketing plan generation module: used to calculate the semantic similarity between the embedded vector and each data in the target knowledge base corresponding to the input category, and combine the semantic similarity with the quantitative indicators of the business performance of the generated marketing plan to generate the optimal marketing plan; Knowledge base optimization module: This module is used to calculate the current business performance quantification index based on the marketing effect after the marketing of the optimal marketing plan, insert marketing plans that meet the admission criteria, and rearrange each plan in the target knowledge base according to the business performance quantification index to obtain the optimized target knowledge base.

8. The AI-based banking marketing optimization system according to claim 7, characterized in that, The knowledge base optimization module includes: Marketing Success Rate Improvement Calculation Submodule: Used to calculate the marketing success rate improvement based on the marketing success rate of the sub-customer group of the optimal marketing plan and the marketing success rate of the sub-customer group of the control group. Customer segment management target improvement calculation submodule: used to calculate the improvement of customer segment management target based on the customer segment marketing target data of the optimal marketing plan, the customer segment marketing target data of the control group, and the importance of each customer segment marketing target; The Business Performance Quantification Submodule is used to calculate quantitative indicators of business performance based on the improvement in marketing success rate, the improvement in customer management goals, and customer grouping. The knowledge base optimization submodule is used to insert marketing plans that meet the admission criteria and rearrange each plan in the target knowledge base according to the quantitative indicators of business performance to obtain the optimized target knowledge base.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the artificial intelligence-based banking marketing optimization method as described in any one of claims 1 to 6.

Images