Intelligent report insight system for bank industry marketing process data analysis

Abstract

The invention relates to the technical field of data management, in particular to an intelligent report insight system for bank industry marketing process data analysis, which comprises the following steps: representing customer behaviors, capturing minute-level product click and other short-term event streams by using a time graph network, and generating dynamic behavior vectors; analyzing a long-term sequence of fund exchange and the like spanning several months by adopting a Transform architecture, and generating a static behavior vector; fusing the two vectors with a customer portrait, inputting the fused vectors into a multi-layer perceptron network, and calculating loan intention intensity by applying semantic projection; meanwhile, the system dynamically groups similar clients by using a clustering algorithm, and quantifies the group fatigue according to the feedback of the groups on recent marketing; a marketing recommendation coefficient is calculated by integrating a loan intention, an account front trend and a group fatigue degree as penalty terms, a business rule base is matched accordingly, an optimal decision covering a touch channel, a product and a verbal skill is generated, client feedback can update the fatigue degree in real time, and a dynamically optimized closed loop is formed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of enterprise data management, in particular to a report intelligent insight system for bank marketing process data analysis. BACKGROUND

[0002] The existing customer marketing analysis system is heavily dependent on static or low-frequency updated data conditions, and cannot effectively capture the real-time demand and intention changes of customers within a specific time window. The system's understanding of customers is often one-sided and lagging, making it difficult to respond to fleeting financial service opportunities revealed by real-time behavior (such as in-app browsing and searching). Most marketing automation systems work based on pre-set, hard-coded rule engines. For example, by setting simple thresholds such as "more than 3 times of browsing a certain product page" or "searching a certain keyword at a set frequency", the marketing action is triggered. This mechanism lacks flexibility and adaptability, and not only cannot identify users with complex behavior patterns but real intentions, but also often generates a large number of false marketing triggers due to simple behaviors (such as accidental clicks and regular browsing), resulting in waste of marketing resources and user harassment. The current system design is generally oriented towards "maximizing marketing reach", focusing on identifying potential positive signals, but almost completely ignores the quantification and management of the negative impact of marketing activities. It cannot automatically reduce marketing frequency or adjust strategies when customers show impatience or resistance, ultimately leading to a decline in customer satisfaction and even customer loss.

[0003] Therefore, a report intelligent insight system for bank marketing process data analysis is proposed. SUMMARY

[0004] The present application aims to provide a report intelligent insight system for bank marketing process data analysis, which solves the problem of low marketing efficiency and poor customer experience caused by inaccurate judgment of customer intentions and general neglect of customer marketing fatigue in existing bank marketing systems. To achieve the above purpose, the present application provides the following technical solutions: A report intelligent insight system for bank marketing process data analysis, comprising: A behavior representation module: using a time graph unit to capture customer short-term behavior event streams and generate dynamic behavior vectors; using a long sequence unit to analyze customer long-term behavior sequences and generate static behavior vectors; inputting the dynamic behavior vectors and the static behavior vectors into a gated cross-attention fusion process to generate a loan intention vector through semantic projection calculation; A loan intention generation module: integrating basic identity attributes, asset information, and credit status to obtain a customer portrait vector, and inputting the customer portrait vector and the loan intention vector into a multi-layer perception network to generate a loan intention strength; The fatigue degree calculation module: the customer portrait vector is one-hot encoded and embedded, spliced with the loan intention vector to form a customer panoramic state vector, a K-Means clustering algorithm is used to define a customer group, and according to the customer panoramic state vector, the group fatigue degree of the recent marketing feedback of the customer group is integrated; The decision recommendation module: a loan trend score is obtained from a loan intention survey questionnaire, a marketing recommendation coefficient is calculated in combination with the loan trend score, the loan intention intensity and the group fatigue degree, and a marketing decision recommendation is generated according to the marketing recommendation coefficient and a business rule library.

[0005] Preferably, the behavior characterization module comprises: the short-term behavior event stream comprises application-in recommended product page clicks and financial tool use records generated by the customer within a minute-level and / or hour-level time window; the customer, product, channel and transaction counterparty are taken as nodes, a heterogeneous graph is constructed by inputting into a time graph network unit, a dynamic behavior vector is obtained by capturing the timing dependency relationship and burst signal between the short-term behavior event stream through internal message passing and node memory updating, and the dynamic behavior vector represents the instant state of the customer in the current network; the long-term behavior sequence comprises large-sum fund transactions, credit product application and repayment records and periodic asset allocation patterns of the customer over a period of months or even years; a long sequence unit based on a Transformer architecture is executed, and the analysis process dynamically evaluates and weights the importance of each behavior in the historical sequence to the loan intention judgment through an internal self-attention mechanism to obtain a static behavior vector.

[0006] Preferably, the behavior characterization module further comprises: the static behavior vector and the dynamic behavior vector are taken as inputs, a gated cross-attention fusion process is executed, the static behavior vector is taken as a query to review the key information in the dynamic behavior vector, and the part of the short-term behavior event stream most related to the long-term pattern is amplified / inhibited through a gating unit to obtain a fused behavior vector; a pre-trained natural language processing model is used to convert the text describing the loan business into a semantic vector, the average value of the semantic vector is calculated, a fixed vector representing the meaning of the loan is synthesized, and the fixed vector is taken as a reference direction for intention judgment; the projection value of the fused behavior vector in the reference direction is calculated as the loan intention vector by semantic projection.

[0007] Preferably, the customer portrait vector generated by the loan intention generation module comprises: the basic identity attributes, the asset information and the credit status are obtained and integrated from multiple sources such as a core transaction system, a credit system, a CRM system and a mobile bank background through ETL extraction, conversion, loading and real-time data synchronization, and a customer portrait vector is constructed with the customer identity as a unique identifier; and the customer portrait vector and the loan intention vector are spliced and input into a multi-layer perception network to calculate and generate a loan intention intensity.

[0008] Preferably, the fatigue degree calculation module comprises: carrying out one-hot encoding and embedding processing on the static portrait of the customer, splicing with the loan intention vector to form a customer panoramic state vector; adopting a K-Means clustering algorithm to perform real-time clustering operation on the customer panoramic state vectors of the whole line; dynamically dividing the customers with similar current behavior patterns, long-term habits and basic attributes into the same customer group; for the customer group, the average click rate, the application conversion rate and the ignore closing rate of the same type of loan marketing information in a preset evaluation period are counted, the fatigue degree score is calculated, and the fatigue degree score is set as the group fatigue degree.

[0009] Preferably, the decision recommendation module comprises: the loan trend score is calculated based on the loan intention research questionnaire analysis of the customer; the loan intention intensity is the output value of the loan intention generation module; the group fatigue degree is the output of the fatigue degree calculation module; the marketing recommendation coefficient is calculated through a preset weighting formula, and the weighting formula gives positive weight to the loan intention intensity and the loan trend score, and gives negative weight to the group fatigue degree. The marketing recommendation coefficient value is compared with a multi-layer threshold value, and the specific conditions in the customer portrait vector are combined to retrieve and match the optimal action strategy in the business rule library to obtain a structured instruction set, wherein the instruction set defines the recommended touch channel, the recommended specific product, the recommended marketing speech template ID and the configurable preferential strategy; the touch channel comprises a call from a customer manager, an APP pop-up window and a short message.

[0010] Preferably, the system further comprises a feedback module for incorporating the marketing acceptance degree into the group fatigue degree, comprising: adopting an event response type dynamic scoring model to calculate an individual dynamic score according to the feedback behavior of the customer to the marketing action, wherein the feedback behavior comprises clicking, applying, ignoring / closing, and if the customer has no feedback within a preset time window, the individual dynamic score will decay at a corresponding rate according to the tolerance of the latest marketing; adopting a weighted average algorithm to apply a weight corresponding to the individual dynamic score to the marketing feedback data of all members in the customer group, and establish a feedback value matrix; when receiving the feedback behavior, the adjustment amount of the individual dynamic score is determined by multiplying the base score by the reference weight corresponding to the current situation in the feedback value matrix, thereby quantitatively calculating the marketing acceptance degree and incorporating the marketing acceptance degree into the group fatigue degree to complete the closed-loop feedback.

[0011] Compared with the prior art, the present application has the following advantages: 1、The application fuses the long-term and short-term behaviors of customers, constructs a comprehensive and dynamic behavior representation, converts the fuzzy intention into an accurate quantitative score by projecting the complex behavior vector to the "loan" semantic benchmark, fundamentally surpasses the traditional rule-based judgment, greatly improves the accuracy and reliability of intention recognition, enables the marketing resources to accurately lock the high-willingness customers, and significantly improves the conversion rate; 2、The application introduces and quantifies the "group fatigue degree" as a decision variable, calculates the fatigue score through dynamic clustering and marketing feedback, uses it as a negative weight in decision-making, forms an "intelligent brake" mechanism, effectively suppresses excessive marketing, avoids customer aversion caused by information bombardment, greatly improves customer experience while pursuing performance, and improves the long-term loyalty of customers; 3、The system constructs an intelligent closed loop from insight to decision to feedback, can comprehensively analyze multi-dimensional information, and automatically generate specific executable marketing instructions; the core is that the real feedback of customers dynamically updates the group fatigue degree, forms a self-optimizing learning loop; this closed loop mechanism enables the marketing strategy to continuously and adaptively evolve, significantly improves the automation and scientificity of decision-making, and reduces the cost of manual calculation. BRIEF DESCRIPTION OF DRAWINGS

[0012] Fig. 1 A step flow chart of the report intelligent insight system for bank marketing process data analysis is provided in the application; Fig. 2 A structural schematic diagram of the report intelligent insight system for bank marketing process data analysis is provided in the application; Fig. 3 A flowchart of the marketing recommendation decision is provided in the application. DETAILED DESCRIPTION

[0013] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments of the application; based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application; Please refer to Figs. 1 to 3 The application provides a report intelligent insight system for bank marketing process data analysis, referring to Fig. 1 A step flow chart of the system is provided, Fig. 2 A structural schematic diagram of the system is provided, and the technical solution is as follows, a report intelligent insight system for bank marketing process data analysis comprises: Behavioral representation module: Uses time-map unit to capture short-term customer behavior event streams to generate dynamic behavior vectors; uses long sequence unit to analyze long-term customer behavior sequences to generate static behavior vectors; inputs the dynamic behavior vectors and the static behavior vectors into a gated cross-attention fusion process, and generates a loan intent vector through semantic projection calculation; Loan Intent Generation Module: Integrates basic identity attributes, asset information and credit status to obtain a customer profile vector, and inputs it along with the loan intent vector into a multilayer perceptron network to generate loan intent strength; Fatigue calculation module: The customer profile vector is one-hot encoded and embedded, and then concatenated with the loan intent vector to form a customer panoramic state vector. The customer group is defined by K-Means clustering algorithm, and the group fatigue of the customer group is integrated based on the recent marketing feedback of the customer group according to the customer panoramic state vector. Decision Recommendation Module: Loan trend score is obtained from loan intention survey questionnaire. The marketing recommendation coefficient is calculated by combining the loan trend score, loan intention strength and group fatigue. Marketing decision recommendations are generated based on the marketing recommendation coefficient and business rule base.

[0014] Example 1: This embodiment mainly introduces how the system's behavior representation module comprehensively captures customers' short-term real-time behavior and long-term historical behavior to generate a loan intent vector. First, the data sources for short-term behavioral event stream capture and dynamic behavior vector generation include: Mobile banking app logs, online banking system logs, counter transaction records, ATM transaction records, SMS interaction records, and customer service call recordings transcribed into text; Time Graph Unit: Employing a 2-layer GAT network with 4 attention heads, outputting 256 dimensions, and using customers, products, channels, and counterparties (merchants) as nodes to construct a heterogeneous graph; Customer A performed the following actions within 10 minutes: logged into the mobile banking app, clicked on the "Deposit Products" page, browsed the details page of the "Current Deposit" product pushed to them, clicked to learn about the details page of the "Fixed Deposit" product; and consulted customer service about "Wealth Management Products" before closing the app.

[0015] Heterogeneous graph construction: Nodes include: Customer A, Mobile Banking App, Deposit Products, Current Deposits, Time Deposits, Wealth Management Products, Fund Zone, Money Market Fund, and a specific money market fund; Edges represent behavioral sequences and timestamps; Further, the time graph network unit updates the state of each node (e.g., customer A's interest in "demand deposit") by aggregating neighbor node information; for example, after customer A clicks on "demand deposit" and then clicks on "fixed deposit", the system will deliver information to update customer A's overall interest in "deposit products"; and node memory update: the time graph network unit maintains the memory state of each node over time; for example, customer A's active behavior on deposit and wealth management products will reflect his immediate high interest in "wealth management" in his dynamic behavior vector; The dynamic behavior vector generation is specifically generating a low-dimensional dense vector (e.g., 256 dimensions) representing customer A's immediate behavior pattern and potential intention within a time window after processing by the time graph network unit; this vector captures the temporal dependence and burst signals between events, for example, consecutive viewing of deposit and fund may indicate that the customer has a demand for capital allocation; Long-term behavior sequence analysis and static behavior vector generation: the data sources mainly include core transaction systems (large-value transfer, salary account), credit systems (loan application, approval, repayment records), CRM systems (customer complaints, suggestions), and asset management systems (wealth management purchase, redemption); Based on the long sequence unit of the Transformer architecture, a 4-layer encoder with 8 attention heads outputs a 512-dimensional vector representing customer A's long-term behavior sequence over the past three years: in March 2021, customer A applied for and was approved for a housing mortgage loan and repaid it on time every month; in July 2021, customer A opened a fund account and regularly invested in an index fund; in January 2022, customer A received a large salary and transferred part of the funds to a demand deposit and used part of the funds for consumption; in May 2022, customer A applied for a credit card and actively used it and repaid it in full; in February 2023, customer A redeemed part of the fund for family education expenses; in August 2023, customer A consulted several times about small consumer loan products but did not apply for them.

[0016] In the Transformer architecture, each event in the long-term behavior sequence (such as "apply for a mortgage" and "fund investment") is considered a node, and the importance of each behavior in the historical sequence to the loan intention judgment is dynamically evaluated and weighted by calculating the attention weight of each event to other events; for example, in judging customer A's loan intention, the event of applying for a mortgage and repaying it on time may have a higher weight because it is directly related to credit behavior; and salary account may also have a higher weight because it reflects the customer's repayment ability; redemption of funds for education expenses may imply future demand for new funds; Extending the "loan intention" to a comprehensive prediction of the customer's "key life cycle events", this is not limited to loans, but also covers a series of major life events such as marriage, childbirth, car purchase, education, and retirement; By fusing the customer's financial behavior (short-term dynamics and long-term statics) with authorized external data (such as public information on social media, consumption trends at partner merchants), the system can predict significant life changes that are about to occur for the customer; for example, if the system monitors that the user has recently frequently browsed baby products, education savings products, and accompanied by large family-related consumption, it predicts not "loan intention", but "newborn family" event; the bank can then push a complete set of services such as family insurance, child-specific savings accounts, and mortgage consultation for larger space housing, upgrading marketing from "single-point promotion" to "scenario-based, consultant-style comprehensive services." Static behavior vector generation includes generating a fixed-length vector (e.g. 512 dimensions) after processing through the Transformer encoder, capturing the long-term behavior patterns, risk preferences, consumption habits, and fund management styles of customer A. This vector is relatively stable and represents the static portrait of the customer.

[0017] Special processing of raw data to convert it into a vector that machines can understand, using time graph network technology to capture the customer's immediate behavior such as clicks and browsing within a short period of time (e.g. within a few hours), generating a dynamic behavior vector; at the same time, it uses the Transformer model to analyze the customer's transaction and repayment history over several years, generating a static behavior vector, providing the core data foundation for accurate judgment of customer intent.

[0018] Further, the gated cross-attention fusion takes the dynamic behavior vector (short-term behavior immediate state) and the static behavior vector (long-term behavior stable pattern) as input; the static behavior vector is used as the query (Query) to examine the key information in the dynamic behavior vector, and the gate unit amplifies / suppresses the part of the short-term behavior event stream that is most relevant to the long-term pattern; Customer C recently frequently browses car loan pages on mobile banking (dynamic behavior vector shows strong car loan intention), but its static behavior vector shows that it has never had a record of large consumption loans in the past, and its asset allocation is conservative; the gate unit will suppress the overly "aggressive" loan intention in the dynamic behavior vector according to the "conservative" characteristics of the static behavior vector, or amplify the signals related to "cautious assessment"; conversely, if the static behavior vector shows that customer C is a high net worth customer with multiple successful loan records, the gate unit will amplify the loan intention in the dynamic behavior vector; The semantic projection calculates the loan intention intensity, including "loan" semantic vector synthesis: with the help of a pre-trained natural language processing model (such as BERT, Word2Vec), texts describing "loan" business, such as "personal consumption loan", "housing mortgage loan", "credit loan", and "enterprise operation loan", are converted into semantic vectors; the average of these semantic vectors is calculated to synthesize a fixed vector (for example, 300 dimensions) representing the meaning of "loan", which serves as the reference direction for intention judgment; The semantic projection uses cosine similarity or dot product to calculate the projection value of the loan intention vector on the reference direction as the loan intention intensity; the higher the projection value, the closer the customer's loan intention vector is to the "loan" semantic vector, and the greater the loan intention intensity; the L2 norm of the loan intention vector is 1, and the L2 norm of the standardized "loan" semantic vector is also 1; if the cosine similarity of the loan intention vector and the "loan" semantic vector is 0.85, the loan intention intensity is 0.85.

[0019] Through a gated cross-attention mechanism, the dynamic vector representing the immediate interest and the static vector representing the long-term habit are fused; then, through the semantic projection technology, the projection value of the fusion vector on the "loan" business direction is calculated, and finally a precise quantitative score is output, which can intuitively reflect the customer's willingness intensity.

[0020] Embodiment two: This embodiment mainly introduces the loan intention generation module, which is responsible for integrating customer basic identity attributes, asset information, and credit status, generating customer portrait vectors, and fusing with loan intention vectors, and finally generating loan intention intensity; The data sources for constructing the customer portrait vector include: core transaction systems (deposit, transfer records), credit systems (historical loans, credit cards, overdue records), CRM systems (occupation, education, family information), and mobile banking backends (device information, login frequency); ETL extraction, transformation, loading, and real-time data synchronization: extraction is to extract data from the above multi-source systems regularly or in real time; transformation is to clean, de-duplicate, and standardize the original data, such as unifying the "gender" field in different systems to "male / female" and unifying the "occupation"; loading is to load the processed data to a unified customer data platform (such as a data warehouse or lake); real-time data synchronization uses a message queue, such as Kafka, to synchronize key operations of customers on mobile banking or online banking in real time, ensuring real-time updating of customer portraits; The customer portrait vector generation takes customer identity as the unique identifier, integrates basic identity attributes (e.g., age, gender, occupation, marital status, region), asset information (e.g., deposit balance, financial product holding, fund market value), credit status (e.g., total debt, overdue record, credit score); after processing such as one-hot encoding and embedding, these features are spliced into a customer portrait vector (e.g., 128 dimensions); The fusion and loan intention intensity generation splices the customer portrait vector and the loan intention vector (e.g., 128+256+512=896 dimensions) and inputs them into a multi-layer perception network MLP; the MLP network includes multiple fully connected layers and activation function ReLU, and the network structure is [896, 256, 64, 1], the activation function is ReLU and Sigmoid; the training parameters use the Adam optimizer with a learning rate of 0.001 to learn the complex nonlinear relationship between different vectors, and finally output a scalar value between 0 and 1, representing the loan intention intensity; the training data uses historical marketing data, and the customers who successfully apply for loans are marked as positive samples (intention intensity close to 1), and the customers who do not apply or are rejected are marked as negative samples (intention intensity close to 0); The customer portrait vector of customer D shows that he is 35 years old, married, a company executive, with an annual income of 500,000, a deposit of 5 million, and no bad credit record; the dynamic behavior vector shows that he has recently frequently viewed overseas housing information; the static behavior vector shows that he has a history of successful large investment behavior; splice the three vectors and input them into MLP; if the output of MLP is 0.92, it means that the loan intention intensity of customer D is very high.

[0021] Integrate scattered data into a structured customer portrait vector, which provides key background information for subsequent intention judgment, ensuring that decisions are based not only on customer behavior but also on their basic situation.

[0022] Example Three: This embodiment mainly introduces the fatigue calculation module, which defines customer groups according to customer portraits, dynamic behaviors and static behaviors, and integrates recent marketing feedback to calculate group fatigue; The customer panoramic state vector generation is to process the customer's static portrait conditions (e.g., high net worth, white collar, student) through one-hot encoding and embedding, and splice them with the loan intention vector to form a customer panoramic state vector (e.g., 128-dimensional portrait conditions + 256-dimensional dynamic behavior + 512-dimensional static behavior = 896 dimensions); K-Means clustering algorithm: K value is determined by elbow rule, typical range is 10-100, real-time clustering operation is performed on the panoramic state vector of the full line customer; Clustering process: the system presets K cluster centers, calculates the distance from the panoramic state vector of each customer to the K cluster centers through iteration, assigns the customer to the nearest cluster, and updates the cluster center until the cluster center no longer changes significantly; Dynamic division: the clustering algorithm dynamically divides the customers into different customer groups according to the current behavior mode, long-term habit and basic attribute similarity of the customers; Assuming K=5; After clustering, customer A is divided into the "high-frequency active investor" group, customer B is divided into the "stable deposit customer" group, customer C is divided into the "potential housing demand customer" group, and customer D is divided into the "small and micro enterprise owner" group; Further, group fatigue degree calculation is performed, for each customer group, the average click rate, application conversion rate and ignore close rate of the same type of loan marketing information in the preset evaluation period (for example, the past 7 days) are counted; Fatigue score calculation formula: Group fatigue degree=1-(w1*average click rate+w2*average application conversion rate-w3*average ignore close rate) Wherein, w1, w2, w3 are weight coefficients, which are set according to business experience, for example, w1=0.4, w2=0.5, w3=0.1; The higher the average click rate and application conversion rate, the lower the fatigue degree; The higher the ignore close rate, the higher the fatigue degree; And in a customer experience priority strategy, w1=0.3, w2=0.3, w3=0.4 can be set, so as to increase the influence of negative feedback on fatigue degree calculation. These weights can be dynamically adjusted according to the marketing goals of different business stages, The "high-frequency active investor" group received 500 financial product marketing pushes in the past 7 days, with an average click rate of 5%, an average application conversion rate of 2%, and an average ignore close rate of 20%; Group fatigue = 1- (0.3*0.05+0.3*0.02-0.4*0.20) = 1.059 (extremely high fatigue, the negative impact of high ignore rate is significantly amplified). Group fatigue = 1- (0.4*0.05+0.5*0.02-0.1*0.20) = 0.99 (high fatigue, because the ignore close rate is high, although the click rate and conversion rate are low); The "potential home buying demand customer" group received 200 mortgage product marketing pushes in the past 7 days, with an average click rate of 15%, an average application conversion rate of 8%, and an average ignore close rate of 5%; Group fatigue = 1- (0.3*0.15+0.3*0.08-0.4*0.05) = 0.951 (moderately high fatigue, even though the click and conversion rates are high, a certain ignore rate keeps it at a high level). Therefore, group fatigue = 1- (0.4*0.15+0.5*0.08-0.1*0.05) = 0.905 (lower fatigue, because the click rate and conversion rate are higher); Quantify the negative impact of marketing, avoid harassing customers, use K-Means clustering algorithm to divide the current customers into the same group as people with similar behavior and attributes; Then the system will calculate the average click rate, ignore rate and other feedback data of this specific group to the same marketing in the near future, and calculate a group fatigue score, which will play a braking role in the final decision.

[0023] Embodiment four: This embodiment mainly introduces the calculation of marketing recommendation coefficient: refer to Fig. 3 The flowchart of marketing recommendation decision proposed by the present application; The loan trend score is calculated by real-time analysis of customer transaction flow and calculation of positive financial change indicators; Examples of indicators are wage growth rate (for example, this month's wage is 10% higher than last month) and continuous growth days of current deposits (for example, current deposits have been growing for 30 consecutive days); The calculation method is to normalize these indicators and then weight and sum to get the loan trend score (for example, 0-100 points); Loan trend score = 0.6*standardized (wage growth) + 0.4*standardized (deposit growth days); Customer E, this month's wage growth rate is 20%, and current deposit has been growing for 45 consecutive days; The loan trend score is calculated as 90 points (high score); Marketing recommendation coefficient formula: Marketing recommendation coefficient = w_intent*loan intent intensity + w_trend*loan trend score - w_fatigue*group fatigue Wherein, w_intention, w_trend, w_fatigue are weight coefficients, which are set according to business objectives; for example, w_intention = 0.5, w_trend = 0.3, w_fatigue = 0.2; Customer F, loan intention intensity 0.8, loan trend score 80 points (0.8 normalized), group fatigue 0.7; marketing recommendation coefficient = 0.5*0.8 + 0.3*0.8 - 0.2*0.7 = 0.5; marketing decision recommendation generation is compared with multi-layer threshold, the marketing recommendation coefficient value is compared with the preset multi-layer threshold (for example, 0.8 or more is "strong recommendation", 0.5-0.8 is "moderate recommendation", 0.3-0.5 is "weak recommendation", and 0.3 or less is not recommended); as shown in Table 2, three example rules of the business rule base, the business rule base is matched with the specific conditions in the customer portrait vector (for example, customer F is "small and micro enterprise owner"), and the optimal action strategy is searched and matched in the business rule base.

[0024] Table 2: Business rule base example The structured instruction set generation is to generate a structured instruction set according to the matched rules, which defines the recommended touch channel, specific product, suggested marketing speech template ID and configurable preferential policy; the marketing recommendation coefficient of customer F is 0.85, and its portrait is "small and micro enterprise owner"; according to the rule base, "small and micro enterprise short-term turnover loan" may be matched, the touch channel is "customer manager call", the marketing speech template ID is "SME_Loan_C", and the preferential policy is "exempt from guarantee".

[0025] For a customer with potential intention but moderate acceptance, the system will not directly recommend "customer manager call", but will automatically arrange a step-by-step warming contact sequence: First step (cultivation): push an article about "how to plan family finance" in the APP.

[0026] Second step (trigger): if the customer reads the article, a low-disturbance financial product recommendation banner will be displayed when he logs in a few days later.

[0027] Third step (conversion): if the customer clicks the banner, the system will push this high-potential lead to the customer manager, and attach the complete interaction path of the customer in advance, to assist artificial precise communication.

[0028] This "journey" will be adjusted in real time according to the feedback of the customer at each step, realizing the mode transition from "one-time sale" to "precise cultivation and conversion", and upgrading the single and static instruction of "generating marketing decision recommendation" to "automatic arrangement of adaptive marketing journey".

[0029] The marketing recommendation coefficient is calculated by the previously calculated intention strength and group fatigue and other indicators. According to the value of the coefficient, the system automatically matches and generates a specific instruction in combination with the business rule library, and clearly defines which channel should be used, which product should be recommended, and even what kind of discount should be attached.

[0030] Embodiment five: This embodiment mainly introduces the feedback module. According to the feedback of customers on marketing decisions, the marketing acceptance degree is quantitatively calculated and is included in the group fatigue. The individual dynamic score calculation adopts an event response type dynamic scoring model. Specifically, the feedback behavior is defined as clicking (positive feedback), applying (strong positive feedback), ignoring (negative feedback), and closing (strong negative feedback). The basic score is set as follows: for example, application = +10 points, click = +2 points, ignore = -5 points, and close = -10 points. Individual dynamic score calculation: customer G has accepted five marketing in the past 30 days: First time (loan A): click (+2 points); Second time (finance B): ignore (-5 points); Third time (loan C): application success (+10 points); Fourth time (credit card D): close (-10 points); Fifth time (loan E): no feedback (decay); The decay mechanism is that if the customer has no feedback within the preset time window (for example, 72 hours), the individual dynamic score will decay according to the tolerance of the latest marketing (for example, the customer manager call tolerance is higher than the APP pop-up window) at the corresponding rate. Customer G has no feedback 72 hours after the fifth marketing. If this marketing is an APP pop-up window (low tolerance), the individual score decay rate is 10% per day; if it is a customer manager call (high tolerance), the decay rate is 20% per day.

[0031] The group fatigue index is updated to establish a feedback value matrix. Each element in the matrix represents the reference weight of different feedback behaviors under a specific marketing situation, as shown in Table 1 below: Table 1: Feedback value matrix in this embodiment Individual dynamic score adjustment: when receiving feedback behavior, the adjustment amount of individual dynamic score is determined by multiplying the base score by the reference weight corresponding to the current situation in the feedback value matrix; customer G applies for loan C; if the base score is +10 and the reference weight of "loan marketing-application" is 0.9, the individual dynamic score adjustment amount is 10*0.9=9 points; weighted average algorithm and aggregation statistics: using the weighted average algorithm, the weight corresponding to the individual dynamic score is applied to the marketing feedback data of all members in the customer group to obtain the group fatigue index through aggregation statistics calculation; Further, the weight function adopts a reverse variant of the Sigmoid function, which reduces the negative impact on the group fatigue when the individual dynamic score is higher than a certain threshold, and vice versa; the updating process is specifically: (1) calculate the latest individual dynamic score of each active customer in the group; (2) according to the Sigmoid reverse variant function, assign a weight to the individual dynamic score of the customer; the weight of high positive score (positive feedback) is low, and the impact on fatigue is small; the weight of low negative score (negative feedback) is high, and the impact on fatigue is large; (3) weighted average of these weighted individual feedback data, combined with the historical group fatigue, to recalculate the updated group fatigue; A certain "potential home buying demand customer" group has an initial fatigue of 0.905; if a large number of customers in this group have positive feedback (click or apply) on mortgage marketing, the individual dynamic score of these positive feedback will obtain a lower weight, and after the overall weighted average, the group fatigue will decrease, for example, to 0.85; on the contrary, if a large number of customers have negative feedback, the group fatigue will increase.

[0032] From "feedback module" and "rule base" to "reinforcement learning optimizer of marketing strategy", the "feedback module" for updating fatigue and the static "business rule base" for matching decision are fused and subverted to build a marketing strategy self-optimization engine based on reinforcement learning, the state is defined as the complete vector of the customer (including behavior, portrait, emotional acceptance, financial health); the action is the set of marketing actions that the system can take (such as pushing a specific product combination through a specific channel at a specific time); the reward is the feedback of the customer, a successful application is a huge positive reward, a click is a small positive reward, and an ignore or "no longer remind" is a negative reward; The core of the system is no longer to execute fixed If-Then rules, but to autonomously explore under what state to take what action to maximize the long-term cumulative reward (i.e., the sum of the bank's long-term business value and customer satisfaction) through continuous "try-feedback-learn"; this approach can discover complex strategies that human experts are difficult to summarize, such as "for a certain specific group of high-knowledge customers, directly recommending high-value products may not be as effective as providing a deep industry analysis report first"; this allows the system to evolve from a passive rule execution tool to a brain that can autonomously learn and create optimal marketing strategies. The above steps give the entire system the ability to learn and evolve, when customers provide specific feedback on marketing, the system will quantify the value of this interaction, update the individual dynamic score of the customer, this score will be adjusted through a weighted algorithm, the overall fatigue index of the group to which the customer belongs, each successful marketing or failed attempt will become an experience for the next decision.

[0033] The system integrates behavior representation, loan intention generation, fatigue calculation, decision recommendation, and feedback modules to build a comprehensive banking marketing intelligence system; it not only can deeply understand the real-time and long-term behavior patterns of customers, accurately identify potential loan intentions, but also can effectively manage customer marketing fatigue, thereby achieving personalized and efficient marketing recommendations, and through continuous feedback learning, the system can continuously optimize marketing strategies, improve customer experience and marketing conversion rate, and bring significant economic and social benefits to banks.

[0034] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements, and variations can be made to these embodiments without departing from the principles and spirit of the present application, the scope of the present application being defined by the appended claims and their equivalents.

Claims

1. A report intelligent insight system for banking marketing process data analysis, characterized in that, Comprise: Behavior characterization module: capture customer short-term behavior event stream using time graph unit, generate dynamic behavior vector; Use long sequence unit to analyze customer long-term behavior sequence, generate static behavior vector; Input the dynamic behavior vector and the static behavior vector into the gated cross attention fusion process, calculate the loan intention vector through semantic projection; Loan intention generation module: integrate basic identity attributes, asset information and credit status to obtain customer portrait vector, input the loan intention vector into the multilayer perception network to generate the loan intention strength; Fatigue calculation module: one-hot encoding and embedding processing are performed on the customer portrait vector, the loan intention vector is spliced to form a customer panoramic state vector, K-Means clustering algorithm is used to define customer groups, and the customer panoramic state vector is integrated with the group fatigue of recent marketing feedback of the customer groups; Decision recommendation module: loan trend score is obtained from loan intention survey questionnaire, marketing recommendation coefficient is calculated by combining loan trend score, loan intention strength and group fatigue; According to the marketing recommendation coefficient and the business rule base, the marketing decision recommendation is generated.

2. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The behavior characterization module comprises: The short-term behavior event stream includes application-in recommended product page clicks, financial tool use records generated by the customer within a minute and / or hour level time window; input into the time graph network unit, take the customer, product, channel and transaction counterparty as the node, build a heterogeneous graph, capture the time sequence dependence relationship and burst signal between the short-term behavior event stream through internal message passing and node memory update, and obtain the dynamic behavior vector, which represents the instantaneous state of the customer in the current network; the long-term behavior sequence includes the customer's historical large amount of money flow, credit product application and repayment records and periodic asset allocation mode over several months or even several years; the analysis process of the long sequence unit based on the Transformer architecture is executed, and the analysis process dynamically evaluates and weights the importance of each behavior in the historical sequence to the loan intention judgment through the internal self-attention mechanism, and obtains the static behavior vector.

3. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The behavior characterization module further comprises: Input the static behavior vector and the dynamic behavior vector, execute a gated cross attention fusion process, take the static behavior vector as the query, review the key information in the dynamic behavior vector, amplify / suppress the part most related to the long-term mode in the short-term behavior event stream through the gating unit, and obtain the fusion behavior vector; with the help of a pre-trained natural language processing model, convert the text describing the loan business into a semantic vector, calculate the average value of the semantic vector, synthesize a fixed vector representing the meaning of the loan, and take the fixed vector as the reference direction of intention judgment; adopt semantic projection to calculate the projection value of the fusion behavior vector in the reference direction as the loan intention vector.

4. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The customer portrait vector generated by the loan intention generation module comprises: From the core transaction system, credit system, CRM system, mobile banking background multiple sources, through ETL extraction, conversion, loading and real-time data synchronization, taking the customer identity as the unique identifier, obtaining and integrating the basic identity attributes, the asset information and the credit status, constructing the customer portrait vector; and inputting the customer portrait vector and the loan intention vector into the multi-layer perception network after splicing, and calculating to generate the loan intention strength.

5. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The fatigue degree calculation module comprises: The static portrait of the customer is one-hot encoded and embedded, spliced with the loan intention vector to form a customer panoramic state vector; the K-Means clustering algorithm is used to perform real-time clustering operation on the customer panoramic state vectors of the whole bank; customers with similar current behavior patterns, long-term habits and basic attributes are dynamically divided into the same customer group; for the customer group, the average click rate, application conversion rate and ignore close rate of the same type of loan marketing information in the preset evaluation period are calculated to obtain the fatigue degree score, and the fatigue degree score is set as the group fatigue degree.

6. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The decision recommendation module comprises: The loan trend score is calculated based on the loan intention survey questionnaire analysis of the customer; the loan intention strength is the output value of the loan intention generation module; the group fatigue degree is the output of the fatigue degree calculation module; The marketing recommendation coefficient is calculated by a preset weighting formula, which gives positive weight to the loan intention strength and the loan trend score, and negative weight to the group fatigue degree; The marketing recommendation coefficient value is compared with the multi-layer threshold value, and the optimal action strategy is retrieved and matched in the business rule library combined with the specific conditions in the customer portrait vector to obtain a structured instruction set, which defines the recommended touch channel, the recommended specific product, the recommended marketing speech template ID and the configurable preferential strategy; the touch channel includes a call from a customer manager, an APP pop-up window and a short message.

7. The report intelligent insight system for banking marketing process data analysis of claim 1, wherein, The system further comprises a feedback module for incorporating the marketing acceptance degree into the group fatigue degree, which specifically comprises: An event response type dynamic scoring model is used to calculate the individual dynamic score according to the feedback behavior of the customer to the marketing action, the feedback behavior including clicking, applying, ignoring / closing, and if the customer has no feedback within the preset time window, the individual dynamic score will decay at a corresponding rate according to the tolerance of the latest marketing; A weighted average algorithm is used to apply the weight corresponding to the individual dynamic score to the marketing feedback data of all members in the customer group to establish a feedback value matrix; when receiving a feedback behavior, the adjustment amount of the individual dynamic score is determined by multiplying the base score by the reference weight corresponding to the current situation in the feedback value matrix, thereby quantitatively calculating the marketing acceptance degree and incorporating it into the group fatigue degree to complete the closed-loop feedback.

Images