A coupon distribution method, electronic device, and storage medium

Abstract

The application provides a coupon distribution method, an electronic device and a storage medium, relates to the technical field of Internet data processing and intelligent marketing, and is used for improving the check-off rate and conversion rate of the pushed coupons. The method comprises the following steps: acquiring multi-source behavior data of a user; constructing user portrait information based on the multi-source behavior data; wherein the user portrait information is used for describing the consumption preference of the user; obtaining a coupon distribution strategy based on the user portrait information, activity configuration information and a large language model; wherein the activity configuration information is used for constraining the activity of the coupon distribution; and executing the coupon distribution strategy.

Description

Technical Field

[0001] This application relates to the fields of Internet data processing and intelligent marketing technology, and in particular to a coupon distribution method, electronic device and storage medium. Background Technology

[0002] With the rapid development of e-commerce and digital marketing, coupons, as an important promotional tool, are widely used in scenarios such as customer acquisition, user activation, repeat purchase increase, and inventory clearance. In order to improve the return on investment of marketing in the fierce market competition, the industry is gradually shifting from the extensive coupon distribution model to a targeted coupon distribution model based on big data analysis.

[0003] Currently, coupon distribution mainly relies on the following methods: One method is to set fixed distribution rules based on human experience, such as issuing the same discount coupon to all users browsing a certain product category. Another method is to use machine learning algorithms to build predictive models, predicting the probability of a user's purchase based on their historical behavioral characteristics (such as the time of their last purchase, purchase frequency, and spending amount), and then selecting eligible users to receive coupons.

[0004] However, the redemption and conversion rates of coupons pushed through the above methods are low. Summary of the Invention

[0005] This application provides a coupon distribution method, electronic device, and storage medium for improving the redemption rate and conversion rate of pushed coupons.

[0006] In a first aspect, this application provides a coupon distribution method, comprising: acquiring multi-source behavioral data of users; constructing user profile information based on the multi-source behavioral data; wherein the user profile information is used to describe the user's consumption preferences; obtaining a coupon distribution strategy based on the user profile information, activity configuration information, and a large language model; wherein the activity configuration information is used to constrain the coupon distribution activity; and executing the coupon distribution strategy.

[0007] The technical solution provided in this application offers at least the following benefits: By acquiring multi-source user behavior data and constructing user profiles to describe user consumption preferences based on this data, a coupon distribution strategy is obtained and executed based on the user profile information, activity configuration information for constraining coupon distribution activities, and a large language model. This method leverages the powerful semantic understanding and reasoning capabilities of the large language model to couple user consumption preferences with complex business constraints, resulting in a coupon distribution strategy with high matching accuracy and reducing the probability of invalid coupon distribution. Furthermore, by accurately matching user needs with coupon resources, it improves user experience and increases the redemption and conversion rates of pushed coupons.

[0008] In some embodiments, constructing user profile information based on multi-source behavioral data includes: standardizing the multi-source behavioral data to obtain a basic feature set; and obtaining user profile information based on the basic feature set.

[0009] In some embodiments, a coupon distribution strategy is obtained based on user profile information, activity configuration information, and a large language model, including: generating context information based on user profile information and activity configuration information; inputting the context information and a prompt word template into the large language model to obtain the coupon distribution strategy; wherein, the prompt word template includes at least one of the following: user feature description, marketing goal description, coupon resource constraints, prohibition policy conditions, and output format constraints.

[0010] In some embodiments, a coupon distribution strategy is obtained based on user profile information, activity configuration information, and a large language model, including: obtaining at least one candidate coupon distribution strategy based on user profile information, activity configuration information, and a large language model; wherein, the candidate coupon distribution strategy includes at least one of the following: coupon type, face value, distribution timing, reach channel, reach copy, and priority parameter; verifying at least one candidate coupon distribution strategy to obtain a verified candidate coupon distribution strategy; scoring the verified candidate coupon distribution strategy according to a reinforcement learning strategy, and selecting the verified candidate coupon distribution strategy with the highest score as the coupon distribution strategy.

[0011] In some embodiments, after executing the coupon distribution strategy, the method further includes: receiving user feedback behavior data; wherein the feedback behavior data includes at least one of the following: claiming behavior, clicking behavior, opening behavior, redemption behavior, ordering behavior, order cancellation behavior, repeat purchase behavior, and churn behavior; and iteratively updating at least one of the parameters of the reinforcement learning strategy, the parameters of the large language model, and the prompt word template of the large language model based on the feedback behavior data.

[0012] In some embodiments, multi-source behavioral data includes at least one of the following: basic user information, browsing behavior data, search behavior data, add-to-cart behavior data, order placement behavior data, payment behavior data, order cancellation behavior data, historical coupon collection records, redemption records, activity participation records, and channel source information.

[0013] In some embodiments, the activity configuration information includes one of the following: coupon resource information, business objective information, and activity constraint information; wherein, the coupon resource information includes at least one of the following: coupon type, face value, minimum spending requirement, applicable scope, validity period, total issuance, activity budget, stacking rules, and frequency limit for claiming; the business objective information includes at least one of the following: new customer acquisition objective, user activation objective, conversion objective, repeat purchase objective, retention objective, inventory consumption objective, and average order value improvement objective; the activity constraint information includes at least one of the following: budget constraint, time constraint, frequency constraint, channel constraint, risk control constraint, inventory constraint, applicable scope constraint, stacking rule constraint, and activity rule constraint.

[0014] In some embodiments, user profile information includes at least one of the following: user identity tags, behavioral tags, intent tags, value tags, and risk tags.

[0015] Secondly, this application provides a coupon distribution device, comprising: a collection unit, a processing unit, and an execution unit; the collection unit is used to acquire multi-source behavioral data of users; the processing unit is used to construct user profile information based on the multi-source behavioral data; wherein the user profile information is used to describe the user's consumption preferences; the processing unit is also used to obtain a coupon distribution strategy based on the user profile information, activity configuration information, and a large language model; wherein the activity configuration information is used to constrain the coupon distribution activity; and the execution unit is used to execute the coupon distribution strategy.

[0016] In some embodiments, the processing unit is specifically used to standardize multi-source behavioral data to obtain a basic feature set; and based on the basic feature set, to obtain user profile information.

[0017] In some embodiments, the processing unit is specifically used to generate context information based on user profile information and activity configuration information; input the context information and prompt word template into a large language model to obtain a coupon distribution strategy.

[0018] In some embodiments, the processing unit is specifically configured to obtain at least one candidate coupon distribution strategy based on user profile information, activity configuration information, and a large language model; wherein the candidate coupon distribution strategy includes at least one of the following: coupon type, face value, distribution timing, reach channel, reach copy, and priority parameter; verify the at least one candidate coupon distribution strategy to obtain a verified candidate coupon distribution strategy; score the verified candidate coupon distribution strategy according to a reinforcement learning strategy, and select the verified candidate coupon distribution strategy with the highest score as the coupon distribution strategy.

[0019] In some embodiments, the processing unit is specifically configured to receive user feedback behavior data; wherein the feedback behavior data includes at least one of the following: claiming behavior, clicking behavior, opening behavior, redemption behavior, ordering behavior, order cancellation behavior, repeat purchase behavior, and churn behavior; and iteratively updating at least one of the parameters of the reinforcement learning strategy, the parameters of the large language model, and the prompt word template of the large language model based on the feedback behavior data.

[0020] Thirdly, this application provides an electronic device comprising: a processor and a memory; the memory storing processor-executable instructions; when the processor is configured to execute the instructions, causing the electronic device to implement the method of the first aspect described above.

[0021] Fourthly, this application provides a computer-readable storage medium comprising: computer software instructions; when the computer software instructions are executed in an electronic device, they cause the electronic device to implement the method described in the first aspect.

[0022] Fifthly, this application provides a computer program product comprising a computer program; when the computer program is run in an electronic device, the electronic device performs the method described in the first aspect.

[0023] The beneficial effects of the second to fifth aspects mentioned above are described in the corresponding description of the first aspect and will not be repeated here. Attached Figure Description

[0024] Figure 1 A schematic diagram illustrating the application environment of a coupon distribution method provided in this application; Figure 2 A flowchart of a coupon distribution method provided in this application; Figure 3 A flowchart of another coupon distribution method provided for this application; Figure 4 A flowchart illustrating yet another coupon distribution method provided in this application; Figure 5 A flowchart illustrating yet another coupon distribution method provided in this application; Figure 6 A flowchart illustrating yet another coupon distribution method provided in this application; Figure 7 A schematic diagram of the composition of a coupon distribution device provided in this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0025] The following is a detailed description of a coupon distribution method, electronic device, and storage medium provided in this application, with reference to the accompanying drawings.

[0026] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0027] The terms "first" and "second," etc., used in the specification and drawings of this application are used to distinguish different objects or to distinguish different treatments of the same object, rather than to describe a specific order of objects.

[0028] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0029] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0030] To facilitate a clear description of the technical solutions of the embodiments of this application, the terms "first" and "second" are used in the embodiments of this application to distinguish the same or similar items with essentially the same function and effect. Those skilled in the art can understand that the terms "first" and "second" are not intended to limit the quantity or execution order.

[0031] In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0032] The technical solution provided in this application can be widely applied to various scenarios involving online marketing and user operations, especially suitable for precision marketing on e-commerce platforms and digital promotions in offline retail stores. Specifically, the method provided in this application can automatically generate and distribute personalized coupon distribution strategies for a large number of users through an intelligent decision-making system during major promotional events or in daily scenarios such as user acquisition, activation, repeat purchases, and inventory clearance.

[0033] With the rapid development of e-commerce and digital marketing, coupons, as a mature promotional tool, are undergoing a transformation in their distribution logic from "universal distribution" to "personalized coupons for each user." Modern marketing systems not only need to process massive amounts of user behavior data but also need to comprehensively consider complex business constraints (such as budget limits and inventory restrictions) and ever-changing operational goals (such as increasing average order value and clearing excess inventory). How to generate optimal coupon distribution strategies in real time under complex constraints has become a pressing technical challenge in the current digital marketing field.

[0034] Currently, coupon distribution strategies primarily rely on the following implementation methods: One type involves setting fixed distribution rules based on human experience, such as issuing the same discount coupon to all users browsing a certain product category, or sending a recall SMS to users who haven't logged in for more than 30 days. The other type utilizes machine learning algorithms (such as logistic regression, collaborative filtering, or gradient boosting trees) to build predictive models. These models are trained based on users' historical statistical characteristics (such as the time of their most recent purchase, purchase frequency, and spending amount) to estimate click-through rates or conversion rates, thereby selecting users with prediction scores higher than a threshold for coupon distribution.

[0035] However, the above methods have significant limitations: distribution methods based on fixed rules lack flexibility and cannot capture users' real-time intentions and personalized preferences, resulting in a large number of coupons being distributed to uninterested users, leading to low redemption and conversion rates. While distribution methods based on machine learning algorithms possess some data-driven capabilities, their feature representation ability is limited, making it difficult to understand complex natural language marketing objectives and generate coupon strategies that include factors such as distribution timing, resulting in low redemption and conversion rates for the pushed coupons.

[0036] To address the aforementioned technical issues, this application provides a coupon distribution method. The method involves acquiring multi-source user behavior data and constructing user profiles to describe user consumption preferences. Then, based on the user profiles, activity configuration information constraining coupon distribution, and a large language model, a coupon distribution strategy is derived and executed. This method leverages the powerful semantic understanding and reasoning capabilities of the large language model to couple user consumption preferences with complex business constraints. The generated coupon distribution strategy has a high matching degree, reducing the probability of invalid coupon distribution. Furthermore, by accurately matching user needs with coupon resources, it can improve user experience and increase the redemption and conversion rates of pushed coupons.

[0037] The embodiments provided in this application will now be described in detail with reference to the accompanying drawings.

[0038] The coupon distribution method provided in this application can be applied to, for example... Figure 1 The application environment shown. For example... Figure 1 As shown, the application environment includes a data acquisition device 10 and a processing device 20. The data acquisition device 10 and the processing device 20 are interconnected.

[0039] In some embodiments, the acquisition device 10 is responsible for sensing and measuring the state or parameters of the target object, capturing real and complete raw data. In this application, the acquisition device 10 is used to collect multi-source behavioral data generated by the target user on the client or server side, and transmit the multi-source behavioral data to the processing device 20 for constructing user profile information.

[0040] Multi-source behavioral data includes, but is not limited to, user basic information, browsing behavior data, search behavior data, add-to-cart behavior data, order placement behavior data, payment behavior data, order cancellation behavior data, historical coupon collection records, redemption records, activity participation records, and channel source information.

[0041] For example, the data acquisition device 10 is equipped with a data access module, which is connected to the business system, membership system, order system, marketing system and log acquisition system, etc. The data access module is used to receive multi-source behavioral data.

[0042] The data access module can connect to external systems via an application programming interface (API), message queue, data collection point tracking, or real-time streaming channel to ensure the real-time performance and integrity of the data access.

[0043] In some embodiments, the data collection device 10 may be a data tracking module, log collection agent, API gateway, or sensor component deployed on a user terminal or server, used to sense and acquire multi-source behavioral data generated by the target user during browsing, searching, transactions, and interactions in real time or in batches. This application does not limit the specific device form or deployment location of the data collection device 10.

[0044] In some embodiments, the processing device 20 is responsible for calculating, analyzing, processing, and making decisions on the data collected by the acquisition device 10. In this application, the processing device 20 is used to receive multi-source behavioral data transmitted by the acquisition device 10, construct user profile information describing user consumption preferences based on the multi-source behavioral data, and then obtain a coupon distribution strategy based on the user profile information, activity configuration information, and a large language model, and execute the coupon distribution strategy.

[0045] For example, the processing device 20 is equipped with a data processing and feature construction module, which is connected to the data access module. The data processing and feature construction module is used to clean, deduplicate, desensitize, standardize and extract features from the collected multi-source behavioral data to generate structured feature data for model inference.

[0046] For example, the processing device 20 is equipped with a user profile building module, which is interconnected with the data processing and feature building module. The user profile building module builds user profile information to describe user consumption preferences based on structured feature data.

[0047] For example, the user profile building module can be implemented by combining rule labels, machine learning labels, and large model semantic labels, so that the constructed user profile information can simultaneously identify users' explicit preferences and implicit needs.

[0048] For example, the processing device 20 is equipped with a coupon resource management module. The coupon resource management module is connected to the marketing backend, coupon center or inventory control system, etc. The coupon resource management module is used to manage the coupon resources that can be distributed, synchronize the availability status of coupon resources in real time, and prevent over-issuance, duplicate issuance or waste of resources.

[0049] The coupon resources include, but are not limited to: coupon type, face value, minimum spending requirement, applicable scope, validity period, total issuance volume, activity budget, stacking rules, frequency of redemption restrictions, and target audience restrictions.

[0050] For example, the processing device 20 is equipped with a large model strategy generation module, which is interconnected with the user profile construction module and the coupon resource management module. The large model strategy generation module generates one or more coupon distribution strategies based on user profile information, coupon resource information, business objective information and activity constraint information, and performs semantic understanding and joint reasoning on this information based on a large language model.

[0051] For example, the processing device 20 is equipped with a rule constraint and risk control verification module, which is interconnected with the large model strategy generation module. The rule constraint and risk control verification module is used to verify the compliance, budgetability, frequency control and risk of the generated coupon distribution strategy.

[0052] For example, the rule constraint and risk control verification module includes at least one of the following: budget constraint submodule, inventory constraint submodule, blacklist filtering submodule, frequency control verification submodule, permission verification submodule, and risk interception submodule.

[0053] The system includes several sub-modules: a budget constraint sub-module to determine if the coupon distribution strategy exceeds the preset activity budget, preventing uncontrolled activity costs; an inventory constraint sub-module to determine if the number of coupons distributed exceeds the current available inventory, preventing over-issuance, empty distribution, or insufficient inventory; a blacklist filtering sub-module to identify whether the users targeted by the coupon distribution strategy are blacklisted, abnormal, or restricted users, thus preventing coupons from being issued to unsuitable users; a frequency control verification sub-module to determine whether the same user has reached the maximum number of coupons issued or the maximum number of times they can be reached within a preset time window, preventing repeated and frequent coupon issuance; an authorization verification sub-module to determine whether the current coupon issuance strategy, channel, or operating entity has the corresponding authorization, preventing unauthorized access or illegal issuance; and a risk interception sub-module to identify whether candidate strategies exhibit abnormal behavior, cheating risks, "wool party" risks, or bulk coupon-buying risks, and to intercept or downgrade high-risk strategies.

[0054] For example, the processing device 20 is equipped with a decision ranking module, which is interconnected with the rule constraint and risk control verification module. The decision ranking module performs a comprehensive score and priority ranking on the verified coupon distribution strategies according to the reinforcement learning strategy to obtain a suitable coupon distribution strategy.

[0055] For example, the processing device 20 is equipped with an intelligent distribution execution module, which is interconnected with the rule constraint and risk control verification module. The intelligent distribution execution module is also interconnected with the user notification system, coupon distribution system, member account system and verification system. The intelligent distribution execution module distributes the final coupons to the target user's account, card wallet, message center, SMS channel, APP push channel or mini-program message channel according to the appropriate coupon distribution strategy.

[0056] In some embodiments, the application environment may further include a feedback device 30, which is interconnected with the processing device 20.

[0057] In some embodiments, the processing device 20 may be a server cluster consisting of multiple servers, a single server, a computer, or a processor or processing chip in a server or computer. This application does not limit the specific device form of the processing device 20.

[0058] In some embodiments, the feedback device 30 is used to collect response data of the target object after performing an operation, and to send the response data back to the processing device 20 for strategy optimization. In this application, the feedback device 30 is used to collect subsequent feedback behavior data of the target user in response to the issued coupon, and to send the collected feedback behavior data to the processing device 20, so that the processing device 20 can iteratively update the parameters of the large language model or the prompt word template of the large language model based on the feedback behavior data.

[0059] The feedback behavior data should include at least one of the following: claiming behavior, clicking behavior, opening behavior, redemption behavior, ordering behavior, order cancellation behavior, repeat purchase behavior, and churn behavior.

[0060] For example, the feedback device 30 is equipped with a feedback learning module, which is interconnected with the intelligent dispatch execution module and the data processing and feature construction module. The feedback learning module can use online learning, incremental learning, reinforcement feedback learning, or periodic offline training methods to update the parameters of the large language model, the parameters of the reinforcement learning strategy, or the prompt word templates of the large language model, thereby forming a closed-loop optimization mechanism.

[0061] In some embodiments, the feedback device 30 may be a behavior listener embedded in the client SDK, a server-side log analysis module, or an ETL task in a data warehouse, used to collect subsequent feedback behavior data of the target user regarding the issued coupons. This application does not limit the specific implementation or deployment form of the feedback device 30.

[0062] In some embodiments, the acquisition device 10, processing device 20 and feedback device 30 described above can be deployed in cloud servers, private servers, hybrid cloud platforms or edge-cloud collaborative environments, microservice architectures or modular service architectures to support high-concurrency request processing, online model inference and continuous feedback learning.

[0063] It should be noted that the system architecture described in the embodiments of this application is for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and does not constitute a limitation on the technical solutions provided in the embodiments of this application. As those skilled in the art will know, with the evolution of system architecture, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0064] See Figure 2 This is a flowchart illustrating a coupon distribution method provided in an embodiment of this application. Figure 2 As shown, the coupon distribution method provided in this application specifically includes the following steps S201-S204.

[0065] S201. Obtain multi-source behavioral data of users.

[0066] In some embodiments, multi-source behavioral data includes at least one of the following: basic user information, browsing behavior data, search behavior data, add-to-cart behavior data, order placement behavior data, payment behavior data, order cancellation behavior data, historical coupon collection records, redemption records, activity participation records, and channel source information.

[0067] For example, multi-source behavioral data of users can be collected from business systems, membership systems, order systems, marketing systems, or log collection systems.

[0068] S202. Construct user profile information based on multi-source behavioral data.

[0069] Among them, user profile information is used to describe users' consumption preferences.

[0070] For example, user profile information is used to describe a user's consumption preferences in dimensions such as activity level, preferences, price sensitivity, consumption level, lifecycle, and conversion tendency.

[0071] In some embodiments, user profile information includes at least one of the following: user identity tags, behavioral tags, intent tags, value tags, and risk tags.

[0072] As one feasible approach, user profile information can be constructed based on multi-source behavioral data, including: standardizing the multi-source behavioral data to obtain a set of basic features; and obtaining user profile information based on the set of basic features.

[0073] For example, multi-source behavioral data can be standardized (e.g., cleaning, deduplication, desensitization, standardization, and feature extraction) to obtain a set of basic features (also known as structured feature data) for inference in large language models.

[0074] The set of basic characteristics includes, but is not limited to: user activity characteristics, spending power characteristics, price sensitivity characteristics, category preference characteristics, channel preference characteristics, lifecycle characteristics, recent conversion characteristics, and churn risk characteristics.

[0075] S203. Based on user profile information, activity configuration information, and large language model, a coupon distribution strategy is obtained.

[0076] The activity configuration information is used to constrain the activities that distribute coupons.

[0077] For example, the activity configuration rules are used to describe budget limits, coupon issuance thresholds, applicable categories, validity periods, frequency control requirements, blacklist restrictions, and stacking restrictions, in order to obtain a set of candidate coupons and a set of business constraints.

[0078] In some embodiments, the activity configuration information includes one of the following: coupon resource information, business objective information, and activity constraint information.

[0079] For example, the coupon resource information includes at least one of the following: coupon type, face value, minimum spending requirement, applicable scope, validity period, total issuance amount, activity budget, stacking rules, and frequency limit for claiming.

[0080] For example, business objective information includes at least one of the following: new customer acquisition objective, user activation objective, conversion objective, repeat purchase objective, retention objective, inventory consumption objective, and average order value improvement objective; For example, the activity constraint information includes at least one of the following: budget constraint, time constraint, frequency constraint, channel constraint, risk control constraint, inventory constraint, scope of application constraint, superimposed rule constraint, and activity rule constraint.

[0081] In some embodiments, such as Figure 3 As shown, step S203 above, based on user profile information, activity configuration information, and the large language model, yields a coupon distribution strategy, specifically including the following steps S301-S302: S301. Generate context information based on user profile information and activity configuration information.

[0082] For example, user profile information, coupon resource information, business objective information, and activity constraint information (such as inventory information and business rule information) are structured and encapsulated to obtain context information.

[0083] The context information can include structured fields or natural language descriptions, which are used to enable the large language model to understand the current marketing goals, constraints, user status, etc.

[0084] As an example, business objective information is used to characterize the business orientation of the current coupon campaign. Business objective information can be configured by marketing operations personnel based on the campaign scenario, or automatically generated by the campaign management module based on a preset campaign template.

[0085] As an example, activity constraint information is used to characterize the restrictions in the current coupon distribution process. Activity constraint information can be generated by the activity configuration module, rule engine, risk control module, or marketing backend.

[0086] S302. Input the context information and prompt word template into the large language model to obtain the coupon distribution strategy.

[0087] The prompt template includes at least one of the following: user characteristic description, marketing objective description, coupon resource constraints, prohibition policy conditions, and output format constraints.

[0088] For example, business rules, activity objectives, operational constraints, and risk control conditions can be converted into prompt word templates for a large language model and input into the large language model along with contextual information to guide the large language model to output a coupon distribution strategy that meets business requirements.

[0089] In some embodiments, such as Figure 4 As shown, step S302 above, which involves inputting contextual information and prompt word templates into the large language model to obtain the coupon distribution strategy, specifically includes the following steps S401-S403: S401. Input the context information and prompt word template into the large language model to obtain at least one candidate coupon distribution strategy.

[0090] The candidate coupon distribution strategy includes at least one of the following: coupon type, face value, distribution timing, distribution channel, outreach copy, distribution frequency, priority, and combination strategy parameters.

[0091] For example, contextual information and prompt word templates are input into a large language model, which then analyzes the relationship between user needs, user behavior characteristics, coupon resource status, and business objectives to generate one or more candidate coupon distribution strategies suitable for the current user or user group.

[0092] S402. Verify at least one candidate coupon distribution strategy to obtain the candidate coupon distribution strategy that passes the verification.

[0093] For example, the rule constraint and risk control verification module performs compliance, budget, frequency control, and risk verification on at least one candidate coupon distribution strategy. If a candidate coupon distribution strategy does not meet the preset business rules, it is removed, modified, downgraded, or rolled back to ensure that the final coupon distribution strategy meets constraints such as budget, frequency, inventory, permissions, and blacklists.

[0094] Among them, compliance, budgetability, frequency control, and risk verification refer to a comprehensive check on whether the candidate coupon distribution strategy meets the requirements of business rules, resource constraints, and security control.

[0095] For example, validating the candidate coupon distribution strategy can be achieved using the following pseudocode:

[0096] S403. Based on the reinforcement learning strategy, score the candidate coupon distribution strategies that have passed the verification, and select the candidate coupon distribution strategy with the highest score as the coupon distribution strategy.

[0097] For example, the validated candidate coupon distribution strategies are scored using reinforcement learning (such as a predefined comprehensive reward function). This scoring process considers at least one factor among estimated conversion probability, estimated return on investment, discount cost, inventory consumption efficiency, predicted user satisfaction, and activity goal alignment. Specifically, weighted summation scoring, learning ranking, multi-objective optimization algorithms, and reinforcement feedback scoring can be used to integrate these factors and output a score for each candidate coupon distribution strategy. Then, the candidate coupon distribution strategy with the highest score that passes validation is selected as the coupon distribution strategy.

[0098] For example, the score can be trained from historical data, or the scoring process can be dynamically updated in conjunction with real-time feedback.

[0099] For example, scoring candidate coupon distribution strategies can be achieved using the following pseudocode:

[0100]

[0101] S204. Implement the coupon distribution strategy.

[0102] For example, coupons are distributed to the target user's account, wallet, or messaging channel according to the coupon distribution strategy, and the distribution results are recorded.

[0103] For example, if a user does not respond within a preset time, a second outreach can be triggered based on reissue or recall rules. This ensures that the implemented coupon distribution strategy can be translated into actionable marketing activities.

[0104] In some embodiments, when executing a coupon distribution strategy, information such as coupon issuance details, coupon identifier, user identifier, issuance time, issuance channel, execution status, and redemption status are recorded for tracking, auditing, and subsequent analysis.

[0105] For example, implementing a coupon distribution strategy can be achieved using the following pseudocode:

[0106]

[0107]

[0108]

[0109]

[0110]

[0111] Based on the above embodiments, by acquiring multi-source user behavior data and constructing user profiles describing user consumption preferences, a coupon distribution strategy is obtained and executed using the user profile information, activity configuration information for constraining coupon distribution activities, and a large language model. This method leverages the powerful semantic understanding and reasoning capabilities of the large language model to couple user consumption preferences with complex business constraints, resulting in a coupon distribution strategy with high matching accuracy, reducing the probability of invalid coupon distribution. Simultaneously, by accurately matching user needs with coupon resources, user experience can be improved, and the redemption and conversion rates of pushed coupons can be increased.

[0112] In some embodiments, after implementing the coupon distribution strategy, the method further includes: receiving user feedback behavior data; and iteratively updating at least one of the parameters of the reinforcement learning strategy, the parameters of the large language model, and the prompt word template of the large language model based on the feedback behavior data.

[0113] The feedback behavior data should include at least one of the following: claiming behavior, clicking behavior, opening behavior, redemption behavior, ordering behavior, order cancellation behavior, repeat purchase behavior, and churn behavior.

[0114] For example, based on feedback behavior data, the constructed user profile information, strategy parameters, scoring weights, or prompts from the large language model can be updated to form a continuously iterative closed-loop optimization mechanism. This mechanism can continuously learn the best coupon distribution patterns under different business scenarios and achieve adaptive optimization of coupon distribution strategies.

[0115] In some embodiments, the data processing and feature construction module can transform unstructured information such as user natural language search content, customer service interaction content, and activity feedback text into semantic features and write them into a feature library or vector library.

[0116] The feature library stores structured feature data to support online inference, offline training, rule computation, and unified feature management. The vector library stores semantic vector data and supports semantic retrieval based on vector similarity, similar user recall, interest matching, and intent recognition.

[0117] For example, feature data in the feature library or vector library can also be used by the feedback learning module to update model parameters, optimize policies, optimize prompt words, or perform incremental training to form a closed-loop optimization mechanism.

[0118] For example, in the process of coupon recommendation, ranking decision or strategy generation, joint analysis can also be performed based on structured features in the feature library or semantic features in the vector library, thereby improving the system's ability to identify user behavior characteristics, consumption intentions and potential needs, and improving the accuracy of recommended coupons, distribution matching degree and redemption conversion effect.

[0119] For example, feedback learning can be implemented using the following pseudocode:

[0120] For example, the entire coupon distribution method described above can be implemented using the following pseudocode:

[0121]

[0122] In some embodiments, such as Figure 5 As shown, when a user enters the system, it's necessary to determine if the user is a new user. For example, if a new user, basic user information is obtained and input into a large language model for analysis. This analysis combines coupon resource information, rule check results, risk control verification results, and optimization results from the reporting feedback learning module to generate a recommended coupon distribution strategy. After the large language model outputs the recommended coupon distribution strategy, it then determines whether to recommend issuing coupons to the user. If the result is yes, the coupon distribution strategy is executed; otherwise, it is not. Alternatively, if the user is not a new user, historical user behavior data is obtained and input into the large language model for analysis. This analysis combines coupon resource information, rule check results, risk control verification results, and optimization results from the reporting feedback learning module to generate a recommended coupon distribution strategy. After the large language model outputs the recommended coupon distribution strategy, it then determines whether to recommend issuing coupons to the user. If the result is yes, the coupon distribution strategy is executed; otherwise, it is not.

[0123] The rule check verifies whether the recommendation results meet the business strategy, activity rules, and distribution constraints. The risk control check identifies abnormal users, abnormal behavior, and potential fraud risks.

[0124] In some embodiments, the large language model can also work in conjunction with traditional recommendation models, ranking models, or reinforcement learning models. Multiple models can operate collaboratively using serial, parallel, or hybrid methods to improve the accuracy, dynamic adaptability, and resource utilization efficiency of coupon distribution strategies.

[0125] Among them, the large language model is used for user intent recognition, semantic understanding, scenario analysis, and generation of candidate coupon distribution strategies. Traditional recommendation models are used to calculate user preference matching degree, predict click-through rate, predict redemption rate, or predict conversion rate. Reinforcement learning models are used to dynamically adjust the distribution strategy based on historical feedback results.

[0126] The coupon distribution method of this application embodiment is described below with reference to a specific example. The specific implementation process of this method is as follows: Figure 6 As shown.

[0127] S601. Obtain multi-source behavioral data of users.

[0128] S602. Standardize the multi-source behavioral data to obtain a set of basic features.

[0129] S603. Based on the aforementioned set of basic features, user profile information is obtained.

[0130] S604. Obtain coupon resource information, business objective information, and activity constraint information.

[0131] S605. Input user profile information, coupon resource information, business objective information and activity constraint information into the large language model to obtain at least one candidate coupon distribution strategy.

[0132] S606. Determine whether the candidate coupon distribution strategy passes the verification.

[0133] For example, if the determination is yes, proceed to step S607; if the determination is no, the process ends.

[0134] S607. Score the candidate coupon distribution strategies and select the coupon distribution strategy with the highest score.

[0135] S608, Execute the coupon distribution strategy with the highest score.

[0136] S609. Obtain user feedback behavior data after implementing the coupon distribution strategy.

[0137] As can be seen, the above mainly describes the solutions provided by the embodiments of this application from a methodological perspective. To achieve the above functions, the embodiments of this application provide corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the modules and algorithm steps of the various examples described in the embodiments disclosed herein, the embodiments of this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed by hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this invention.

[0138] This application embodiment can divide the coupon distribution device into functional modules according to the above method example. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. Optionally, the module division in this application embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0139] In some embodiments, this application also provides a coupon distribution device. The coupon distribution device may include one or more functional modules for implementing the coupon distribution method of the above method embodiments.

[0140] For example, Figure 7 This is a schematic diagram illustrating the composition of a coupon distribution device provided in an embodiment of this application. Figure 7 As shown, the coupon distribution device 700 includes: a data acquisition unit 701, a processing unit 702, and an execution unit 703; the data acquisition unit 701 is used to acquire multi-source behavioral data of users; the processing unit 702 is used to construct user profile information based on the multi-source behavioral data; wherein, the user profile information is used to describe the user's consumption preferences; the processing unit 702 is also used to obtain a coupon distribution strategy based on the user profile information, activity configuration information, and a large language model; wherein, the activity configuration information is used to constrain the coupon distribution activity; the execution unit 703 is used to execute the coupon distribution strategy.

[0141] In some embodiments, the processing unit 702 is specifically used to standardize multi-source behavioral data to obtain a basic feature set; and based on the basic feature set, to obtain user profile information.

[0142] In some embodiments, the processing unit 702 is specifically used to generate context information based on user profile information and activity configuration information; input the context information and prompt word template into a large language model to obtain a coupon distribution strategy.

[0143] In some embodiments, the processing unit 702 is specifically configured to obtain at least one candidate coupon distribution strategy based on user profile information, activity configuration information, and a large language model; wherein, the candidate coupon distribution strategy includes at least one of the following: coupon type, face value, distribution timing, reach channel, reach copy, and priority parameter; verify the at least one candidate coupon distribution strategy to obtain a verified candidate coupon distribution strategy; score the verified candidate coupon distribution strategy according to a reinforcement learning strategy, and select the verified candidate coupon distribution strategy with the highest score as the coupon distribution strategy.

[0144] In some embodiments, the processing unit 702 is specifically used to receive user feedback behavior data; wherein the feedback behavior data includes at least one of the following: claiming behavior, clicking behavior, opening behavior, redemption behavior, ordering behavior, order cancellation behavior, repeat purchase behavior, and churn behavior; and iteratively updating at least one of the parameters of the reinforcement learning strategy, the parameters of the large language model, and the prompt word template of the large language model based on the feedback behavior data.

[0145] In the case of implementing the functions of the integrated modules described above in hardware, this embodiment of the invention provides a possible structural schematic diagram of the electronic device involved in the above embodiments. For example... Figure 8 As shown, the electronic device 900 includes: a processor 902, a communication interface 903, and a bus 904. Optionally, the electronic device 900 may also include a memory 901.

[0146] Processor 902 may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 902 may be a central processing unit, a general-purpose processor, a digital signal processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It may implement or execute various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this application. Processor 902 may also be a combination that implements computing functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0147] The communication interface 903 is used to connect to other devices via a communication network. This communication network can be Ethernet, wireless access network, wireless local area network (WLAN), etc.

[0148] The memory 901 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto.

[0149] In one possible implementation, the memory 901 can exist independently of the processor 902. The memory 901 can be connected to the processor 902 via a bus 904 and is used to store instructions or program code. When the processor 902 calls and executes the instructions or program code stored in the memory 901, it can implement the coupon distribution method provided in this embodiment of the invention.

[0150] In another possible implementation, the memory 901 can also be integrated with the processor 902.

[0151] The 904 bus can be an extended industry standard architecture (EISA) bus, etc. The 904 bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0152] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the coupon distribution device can be divided into different functional modules to complete all or part of the functions described above.

[0153] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by computer instructions instructing related hardware. The program can be stored in the aforementioned computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be any of the foregoing embodiments or memory. The aforementioned computer-readable storage medium can also be an external storage device for the aforementioned coupon distribution device, such as a plug-in hard drive, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the aforementioned coupon distribution device. Further, the aforementioned computer-readable storage medium can include both internal storage units of the aforementioned coupon distribution device and external storage devices. The aforementioned computer-readable storage medium is used to store the aforementioned computer program and other programs and data required by the aforementioned coupon distribution device. The aforementioned computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0154] This application also provides a computer program product, which includes a computer program that, when run on a computer, causes the computer to execute any of the coupon distribution methods provided in the above embodiments.

[0155] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for distributing coupons, characterized in that, The method includes: Acquire multi-source user behavior data; Based on the multi-source behavioral data, user profile information is constructed; wherein, the user profile information is used to describe the user's consumption preferences; Based on the user profile information, activity configuration information, and large language model, a coupon distribution strategy is obtained; wherein, the activity configuration information is used to constrain the coupon distribution activity. Execute the coupon distribution strategy.

2. The method according to claim 1, characterized in that, The process of constructing user profile information based on the multi-source behavioral data includes: The multi-source behavioral data is standardized to obtain a basic feature set; Based on the aforementioned set of basic features, user profile information is obtained.

3. The method according to claim 1, characterized in that, The coupon distribution strategy, derived based on the user profile information, activity configuration information, and large language model, includes: Context information is generated based on the user profile information and activity configuration information; The context information and prompt word template are input into the large language model to obtain the coupon distribution strategy; wherein, the prompt word template includes at least one of the following: user feature description, marketing goal description, coupon resource constraints, prohibition strategy conditions, and output format constraints.

4. The method according to claim 1, characterized in that, The step of inputting the context information and prompt word template into a large language model to obtain a coupon distribution strategy includes: The context information and prompt word template are input into the large language model to obtain at least one candidate coupon distribution strategy; wherein, the candidate coupon distribution strategy includes at least one of the following: coupon type, face value, distribution timing, distribution channel, outreach copy, distribution frequency, priority, and combination strategy parameters; Verify at least one of the candidate coupon distribution strategies to obtain the candidate coupon distribution strategies that pass the verification; The candidate coupon distribution strategies that pass the verification are scored according to the reinforcement learning strategy, and the candidate coupon distribution strategy with the highest score is selected as the coupon distribution strategy.

5. The method according to claim 1, characterized in that, After executing the coupon distribution strategy, the method further includes: Receive user feedback behavior data; wherein, the feedback behavior data includes at least one of the following: claim behavior, click behavior, open behavior, redemption behavior, order placement behavior, order cancellation behavior, repeat purchase behavior, and churn behavior; Based on the feedback behavior data, at least one of the parameters of the reinforcement learning strategy, the parameters of the large language model, and the prompt word template of the large language model is iteratively updated.

6. The method according to claim 1, characterized in that, The multi-source behavioral data includes at least one of the following: User basic information, browsing behavior data, search behavior data, add-to-cart behavior data, order placement behavior data, payment behavior data, order cancellation behavior data, historical coupon redemption records, redemption records, activity participation records, and channel source information.

7. The method according to claim 1, characterized in that, The activity configuration information includes one of the following: Coupon resource information, business objective information, and activity constraint information; The coupon resource information includes at least one of the following: coupon type, face value, minimum spending requirement, applicable scope, validity period, total issuance volume, activity budget, stacking rules, and frequency limit for claiming coupons; The business objective information includes at least one of the following: new customer acquisition objective, user activation objective, conversion objective, repeat purchase objective, retention objective, inventory consumption objective, and average order value improvement objective; The activity constraint information includes at least one of the following: budget constraint, time constraint, frequency constraint, channel constraint, risk control constraint, inventory constraint, scope of application constraint, superimposed rule constraint, and activity rule constraint.

8. The method according to claim 1, characterized in that, The user profile information includes at least one of the following: User identity tags, behavior tags, intent tags, value tags, and risk tags.

9. An electronic device, characterized in that, The device includes a processor and a memory, the processor being coupled to the memory; the memory is used to store computer instructions, which are loaded and executed by the processor to enable the computer device to implement the coupon distribution method as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes computer-executable instructions that, when executed on a computer, cause the computer to perform the coupon distribution method according to any one of claims 1 to 8.

Images