BC integrated marketing method based on one object and one code

Abstract

The invention provides a one-object-one-code-based bC integrated marketing method, and particularly relates to the field of bC integrated marketing, and the method comprises the steps: introducing a Neure-Symbolic reasoning method into a one-object-one-code bC integrated marketing system, carrying out the dynamic reasoning, and binding a b-end control strategy and a C-end excitation path which are adaptive to the current attribute of a commodity, a sales scene and a user tag, thereby achieving the one-object-one-code-based bC integrated marketing. A one-to-one mapping relation between an entity code and a digital twin code is constructed, and adaptive evolution and remote updating of a strategy are realized based on a code scanning behavior in a commodity circulation process, so that the problems that the strategy is fixed, the response is rigid, the behavior data utilization rate is low and a B end is difficult to perceive feedback of a C end in real time in an existing one-object-one-code system are solved; the strategy personalization of the commodity granularity level, the logic dynamic updating driven by the code scanning behavior and the visual feedback of the code scanning heat are realized, and the bC linkage efficiency and the accuracy of the marketing strategy are greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of bC integrated marketing, and in particular to a bC integrated marketing method based on one item one code. Background Art

[0002] With the rapid development of the Internet of Things (IoT) and digital marketing, one-to-one code technology has been widely applied in areas such as product traceability, anti-counterfeiting, and user interaction. In existing integrated business-to-consumer (B2C) marketing systems, by assigning each product a unique physical code, fundamental data on product circulation and consumer behavior is collected. Combining big data analysis with user profiling, merchants can develop marketing strategies, such as scan code incentives and coupon distribution, to promote consumer engagement and boost sales results. Furthermore, some systems have begun analyzing product sales channels and user feedback to adjust marketing routes and control warehousing strategies, achieving coordinated management of both the B2C and C2C sides.

[0003] However, existing technologies mostly rely on static or preset policy rules, lacking a deep understanding of and real-time response capabilities to the multi-dimensional dynamic behavior of goods during circulation. Traditional methods make it difficult to achieve synchronous updates of product codes and digital twins, resulting in a disconnect between scanning behavior feedback and marketing strategies, delayed strategy adjustments, and impacting real-time optimization and precise delivery of marketing results. In addition, the warehousing end lacks support for real-time visualization of consumer scanning thermal data, making it difficult to fully control the status of commodity circulation and market reactions, limiting the B-end's ability to fine-tune management of C-end behavior. In response to the above shortcomings, there is an urgent need for an innovative marketing method that can dynamically iterate product code strategies and achieve real-time interaction between scanning behaviors and digital twins, so as to enhance the intelligence and response speed of the B-C integrated marketing system. Summary of the Invention

[0004] The present invention proposes a bC integrated marketing method based on one item and one code, aiming to solve the problems in the existing bC integrated marketing system of disconnection between product scanning behavior feedback and marketing strategy, lagging strategy adjustment, and lack of real-time visualization of scanning heat at the warehousing end. Through the Neural-Symbolic fusion reasoning method based on one item and one code, dynamic binding and strategy iteration of product code and digital twin code are realized, scanning behavior events are collected and analyzed in real time, and the control strategy of the b-end and the incentive path of the c-end are dynamically adjusted, ultimately realizing intelligent marketing management with collaborative optimization of the b-end and c-end.

[0005] Among them, a bC integrated marketing method based on one item one code includes the following steps: S1. Generate a feature vector for the product based on its attribute parameters, sales scenario, and user tags. Combined with Neural-Symbolic methods, the system infers the b / c strategy combination bound to the feature vector. Jointly determine the b / c control strategy and c / c incentive path to be adopted for the product at the current stage, and output the b / c strategy combination structure. Specifically, the above steps construct a structured feature model for the product based on its attributes (such as category, SKU, expiration date), sales scenario (such as online promotion, offline counter, regional distribution), and user tags (such as user profile, activity level, and consumption habits), generating a high-dimensional feature vector. This vector is fed into the Neural-Symbolic inference engine. The neural network component first performs pattern recognition and similarity learning on historical policy performance data to initially obtain a set of candidate feasible policy combinations. The symbolic logic component then performs rule filtering and logical consistency verification on the candidate policies to ensure that the policies are consistent with the current product status and marketing policy. Finally, the Neural and Symbolic components jointly make decisions and output a complete business-to-consumer (BC) policy combination structure that includes both the business-side control strategy and the consumer-side incentive path, achieving personalized policy matching for the current product status.

[0006] S2. Based on the inferred bC strategy combination, a physical code is generated for the product. A digital twin code containing the response logic is also generated for each physical code. The physical code and digital twin code are bound together using the product's unique code, forming a one-to-one code pair mapping relationship. Specifically, the above steps generate a "physical code" specifically for the commodity entity based on the bC strategy combination inferred in S1. This code contains static identification and policy-directed information, which is used for the actual deployment of the commodity in the logistics and sales links. At the same time, a "digital twin code" is generated synchronously. Its structure includes response logic, incentive control parameters, and status update identification for scanning events. It is a virtual mapping model of commodity behavior. The physical code and the digital twin code are bound through the commodity's unique coding field to ensure that any C-end scanning behavior can be accurately mapped to the corresponding digital twin logic, thereby forming a one-to-one code pair mapping relationship, realizing a two-way closed-loop mapping and management of each commodity in the real world and virtual system.

[0007] S3. Collect the timestamp, spatial coordinates, and response link information from the product scanning behavior event and input it into the Neural-Symbolic structure for behavior analysis and policy adjustment. The new policy combination is output and the logical content of the original digital twin code is updated based on the new policy combination. Specifically, after the product is put on the market, the user's scanning behavior forms event data, including the scanning timestamp, the spatial coordinates of the scanning location, and the response link path actually triggered after the scanning. This information is accurately collected according to the mapping relationship between the physical code and the digital twin code, forming a spatiotemporal data stream of the product scanning behavior. This behavioral data is input into the Neural-Symbolic structure. The neural part is used to identify behavioral feature patterns and trend changes, and the symbolic part evaluates the actual adaptability and potential failure risk of the current binding strategy. If the reasoning result shows that the current strategy combination is no longer applicable under the current behavior, the system will generate a new bC strategy combination based on the reasoning result, and update the response logic and incentive parameters in the corresponding digital twin code through the remote interface to achieve strategy evolution; if the strategy does not trigger the evolution condition, it will continue to collect scanning behavior, maintain the original strategy structure, and form a dynamic loop update mechanism.

[0008] S4. Based on the scan events recorded in the updated digital twin code, coordinate clustering statistics are performed on the C-end scan location field to generate a scan density distribution map. The thermal distribution results are then called by product number on the warehouse side to visualize the B-end warehouse.

[0009] Specifically, the updated digital twin code retains structured scanning event information, especially the geographic coordinate field of the user's scanned code. The system first standardizes these coordinate data and removes outliers, and then uses clustering algorithms (such as DBSCAN or Mean-Shift) to perform spatial clustering analysis to identify C-end scanning hotspots and their density distribution. The analysis results are generated in the form of a scan density distribution map to form a heat map data structure, and are classified and stored according to the unique product number. B-end warehouse management personnel can call this heat map to visually monitor the active scanning areas of different products in the warehouse system to assist in subsequent business decisions such as warehouse division, replenishment, and strategy adjustment, and build a dynamic perception closed loop that reflects C-end behavior back to B-end management.

[0010] Furthermore, step S1 dynamically generates the optimal bC strategy combination for each product by inputting product attributes, sales scenarios, and user tags into a Neural-Symbolic architecture for joint inference. This breaks the traditional static approach of "fixed strategy + batch coding" and achieves "on-demand strategy, code-driven strategy." This improves the fit between products and user scenarios, enhances the flexibility of strategy response, and enhances the effectiveness of targeted delivery. Step S2, through a one-to-one binding mechanism between physical codes and digital twin codes, ensures that each product not only has a unique physical identification code but also possesses logical response capabilities and the ability to track strategy status. Scanning is no longer just a trigger for information display; it becomes a behavioral entry point with real-time feedback and updates, establishing a digital closed loop for product behavior management, improving the utilization rate of scan events and the quality of strategy responses. Step S3 collects real-world behavioral data, such as scan time, spatial location, and response links, and inputs it into the Neural-Symbolic inference engine to assess strategy adaptability and failure risk. The system then automatically identifies the strategy's lifecycle status. When the current strategy is no longer effective, it can be automatically evolved and remotely updated, effectively avoiding "strategy stagnation" or "incentive waste" and improving strategy efficiency and lifecycle management. In step S4, by clustering and constructing a heat map of the scanned geographic coordinate data, the B-end warehousing system can view the scanning activity of products in various regions in real time, thereby assisting in refined management such as regional delivery optimization, replenishment priority adjustment, and promotion rhythm arrangement. Compared with the traditional model of relying solely on sales data for warehousing decisions, this improves the B-end visualization and utilization value of C-end behavioral data and builds a perception-driven warehousing response mechanism.

[0011] The beneficial effects of the invention are: The present invention introduces the Neural-Symbolic reasoning method into the one-item-one-code bC integrated marketing system, dynamically infers and binds the b-end control strategy and the c-end incentive path adapted to the current attributes of the product, sales scenarios and user tags, constructs a one-to-one mapping relationship between the physical code and the digital twin code, and realizes the adaptive evolution and remote update of the strategy based on the scanning behavior during the circulation of the product, thereby solving the problems of fixed strategy, rigid response, low utilization rate of behavioral data and difficulty for the b-end to perceive the c-end feedback in real time in the existing one-item-one-code system, realizing the personalization of strategies at the product granularity level, dynamic update of logic driven by scanning behavior and visual feedback of scanning heat, greatly improving the bC linkage efficiency and the accuracy of marketing strategies. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 A flowchart of a bC integrated marketing method based on one item one code provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following.

[0014] In order to make the objectives, technical solutions, and advantages of the present invention more clearly understood, the present invention is further described in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present invention and are not intended to limit the present invention. That is, the embodiments described herein are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and illustrated in the drawings herein may be arranged and designed in various different configurations.

[0015] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making any creative work shall fall within the scope of protection of the present invention. It should be noted that relational terms such as "first" and "second" are merely used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations.

[0016] Furthermore, the terms "comprises," "comprising," or any other variations thereof are intended to cover a non-exclusive inclusion such that a process, method, article, or machine that comprises a list of elements includes not only those elements but also other elements not expressly listed or inherent to such process, method, article, or machine. In the absence of more limitations, an element defined by the phrase "comprising a..." does not preclude the presence of additional identical elements in the process, method, article, or machine that comprises the element.

[0017] The features and performance of the present invention are further described in detail below with reference to the embodiments.

[0018] Example 1 like Figure 1 , a bC integrated marketing method based on one item one code, including the following steps: S1. Generate a feature vector for the product based on its attribute parameters, sales scenario, and user tags. Combined with the Neural-Symbolic method, the system infers the associated b / C strategy combination. It then makes a joint decision on the b / C control strategy and the customer incentive path that should be adopted for the product at the current stage, outputting the b / C strategy combination structure. S2. Based on the inferred bC strategy combination, a physical code is generated for the product. A digital twin code containing the response logic is also generated for each physical code. The physical code and digital twin code are bound together using the product's unique code, forming a one-to-one code pair mapping relationship. S3. Collect the timestamp, spatial coordinates, and response link information from the product scanning behavior event and input it into the Neural-Symbolic structure for behavior analysis and policy adjustment. The new policy combination is output and the logical content of the original digital twin code is updated based on the new policy combination. S4. Based on the scan events recorded in the updated digital twin code, coordinate clustering statistics are performed on the C-end scan location field to generate a scan density distribution map. The thermal distribution results are then called by product number on the warehouse side to visualize the B-end warehouse.

[0019] In steps S1 and S3, the neural network automatically extracts deep-level features from the multidimensional attributes of products and scanning behavior data. Specifically, the network architecture is based on a multi-layer perceptron (MLP) or combined with time series models (such as LSTM and Transformer) to adapt to the nonlinear relationships between product attributes and the spatiotemporal dynamics of scanning behavior. The input layer receives structured, standardized feature vectors (including product category, SKU, sales scenario, user tags, etc.). Through nonlinear transformations in the hidden layer, it extracts latent feature representations that are closely related to the effectiveness of the bC strategy, enhancing the model's adaptability to complex environments. Furthermore, in step S1022, the neural network encoder maps the standardized features of the current product into a high-dimensional feature space and uniformly encodes the performance characteristics of strategy combinations stored in the historical database. Using the weighted cosine similarity metric introduced during training, the model can quickly and accurately identify the strategy combinations that best match the current product and have shown good historical performance, forming a set of potentially effective candidates. In step S3023, the neural network acts as an encoder, converting the multidimensional scanning behavior feature matrix (time interval, path offset, response link) into a low-dimensional feedback vector, capturing the spatiotemporal patterns of the scanning behavior and the response characteristics. This vector, combined with the symbolic logic module, participates in reasoning, leveraging the generalization and learning capabilities of the neural network while incorporating the logical rigor of business rules.

[0020] Furthermore, in step S1, a neural-symbolic approach is used to combine the multidimensional attributes of products with business rules to achieve joint reasoning and decision-making for product-to-consumer (B / C) strategy combinations. First, the neural network module performs multi-layer nonlinear mapping on the product's structured attribute parameters (including category, SKU, sales scenario, user tags, etc.), automatically extracting high-dimensional latent feature representations that capture the complex intrinsic relationship between product attributes and the sales environment. Subsequently, the symbolic reasoning module formalizes the policy constraints and logical relationships defined in the business rule library into symbolic expressions. Compliance verification and logical screening are performed based on the candidate strategy set extracted by the neural network. The two complement each other: the neural network performs data-driven feature abstraction and similarity matching, while symbolic reasoning ensures the legitimacy of the strategy combination and consistency with business logic. Finally, a multi-objective optimization algorithm is used to make a joint decision on the strategy set selected by symbolic reasoning, outputting the optimal combination of B-side behavior control strategies and C-side incentive paths that meets the needs of the current product stage. This Neural-Symbolic fusion mechanism combines the adaptive capabilities of deep learning with the interpretability of symbolic logic, enabling dynamic reasoning of complex, multidimensional product characteristics and business rules, effectively improving the accuracy and compliance of policy combinations. In step S3, the Neural-Symbolic architecture focuses on dynamic analysis of scanning behavior feedback and policy adjustments. The neural network encoder receives a constructed scanning behavior feature matrix, which contains the time interval, spatial path offset, and response link information of the scanning event. Using a deep neural network, it extracts a low-dimensional feedback vector of the scanning behavior, capturing the spatiotemporal dynamic patterns of scanning behavior and user response characteristics. The symbolic logic module semantically interprets the feedback vector based on business rules, calculates a symbolic logic score, and assesses the adaptability of the current policy combination to actual scanning behavior and its potential failure risk. These two components combine to form the Neural-Symbolic inference engine for feedback analysis, enabling accurate identification of scanning behavior and real-time assessment of policy effectiveness. When the symbolic logic score and neural network feedback jointly indicate that the existing strategy is invalid or mismatched, the inference engine triggers strategy evolution, dynamically generates a new bC strategy combination, and instantly updates the response logic and incentive parameters in the digital twin code through a remote interface. This mechanism organically integrates deep learning driven by code scanning behavior data with symbolic logic reasoning, enabling intelligent iteration of the digital twin code and dynamic adaptive adjustment of the bC strategy.

[0021] Furthermore, the step S1 specifically includes the following sub-steps: S101. Structuralize product attribute data based on product category, SKU, target sales region, and user profile, construct a standardized feature vector, and input this standardized feature vector into the Neural-Symbolic fusion inference framework. S102. Calling the trained neural network model to perform similarity matching between the performance characteristics of the product and the bC strategy combinations in the historical database to identify potential effective strategy combination candidates, wherein the bC strategy combination includes the b-side control strategy and the c-side incentive path; S103. Based on the policy constraints set in the business rule library, the symbolic reasoning engine performs compliance screening and logical verification on the candidate policy combinations output by the neural network, eliminating policy paths that do not comply with the rules; S104. Based on the strategy set filtered by symbolic reasoning, a multi-objective optimization method is used for joint decision-making to determine and output the optimal strategy combination of the B-end behavior control strategy and the C-end incentive path that should be bound to the product at the current stage.

[0022] Specifically, by modeling the structured attributes of products, a standardized feature vector is established as the inference input. First, a trained neural network is used to extract the degree of match between product characteristics and historical strategies, and candidate strategy combinations with high similarity are screened from the historical database. The candidate results are then submitted to the symbolic reasoning logic structure for constraint judgment and rule verification. Pre-set business logic expressions are used to filter out logically incoherent or conflicting strategy paths. Finally, an optimization algorithm is used to conduct a joint objective evaluation of strategies that meet the constraints (such as balancing ROI maximization and user activity). The result is the optimal combination of business-side control and customer-side incentive paths for the current stage of the product's lifecycle, achieving a dynamic strategy matching process that is jointly driven by data and symbolic decision-making.

[0023] Furthermore, in step S102, the B-end control strategy includes warehouse scheduling, shelf layout control, inventory rhythm, batch strategy, code life cycle setting and product permission configuration, and the C-end incentive path includes code scanning reward parameters, behavior chain guidance, incentive presentation form, user verification mechanism, cumulative behavior tracking, sharing fission path and incentive dynamic update mechanism.

[0024] Furthermore, the step S102 specifically further includes the following sub-steps: S1021 reads the stored strategy combination vector from the historical bC strategy database, each strategy combination vector corresponds to the performance characteristics of a strategy combination; S1022. Calculate the weighted cosine similarity between the normalized feature vector of the current product and each historical strategy combination vector; S1023. Using a preset similarity threshold as a screening criterion, select the strategy combinations that are most relevant to the current product and have good historical performance to form a candidate strategy set.

[0025] Specifically, in S1021, the system first accesses the historical bC strategy database through an interface. Each record in this database contains a complete set of strategy combinations and their performance evaluation results in historical applications. These strategy combinations are typically preprocessed into fixed-length vectors, with each vector dimension corresponding to a specific strategy execution parameter or behavioral performance indicator, such as code scanning response latency, user incentive acceptance, inventory turnover cycle, and consumer-end fission propagation range. This represents the overall performance of the strategy combination in real-world distribution scenarios. The system sequentially reads these strategy combination vectors through the data processing module and caches them in memory, serving as a reference for subsequent matching operations. In S1022, the system uses the standardized feature vector generated for the current item in S101 as the target vector and invokes the matching engine to perform similarity calculations against each of the historical strategy combination vectors. During this calculation, the system uses a feature importance model instead of a general cosine formula. The system learns the contribution of each dimension to matching quality through model learning and weights the feature dimensions differently when calculating similarity, giving higher-weighted dimensions a greater influence in the matching process. This improves the accuracy of strategy recognition and the stability of behavior prediction. In S1023, the system performs threshold judgments based on the similarity results. The system sets a set of floating similarity thresholds to automatically adapt to the behavioral differences in strategy reuse across different product categories. When a strategy combination's matching score exceeds the threshold and its historical execution performance meets the system's built-in performance optimization criteria, the strategy is included in the candidate set, along with its source and matching score as input to the subsequent rule engine. This ensures that the candidate strategies are not only structurally similar but also have traceability and adaptability to actual performance. This process enhances the effectiveness of strategy screening through the synergy of neural matching and historical evidence, and is a key embodiment of the "experience-driven" module in the Neural-Symbolic architecture.

[0026] Furthermore, the step S103 specifically further includes the following sub-steps: S1031. Load the current business rule library involving constraints on the B-side control and C-side incentive path; S1032. For each candidate strategy combination, use the symbolic reasoning engine to apply business rules in sequence, map the strategy combination parameters to the rule expression and perform logical verification; S1033. Mark the policy combinations that fully meet all rule conditions as compliant, eliminate the non-compliant policy combinations, and output the policy set after symbolic reasoning screening.

[0027] Specifically, in S1031, the system dynamically loads the rule set effective for the current operational phase from a pre-configured business rule library. This rule set covers business-side control dimensions (such as inventory limits, transfer restrictions, shelf configuration requirements, and warehouse priorities) as well as customer-side incentive path constraints (such as user participation thresholds, incentive presentation methods, guidance link length limits, and platform compliance requirements). These rules are stored as formal logic expressions and have callable interfaces to facilitate subsequent inference and invocation. In S1032, the system maps each candidate policy combination screened in S102, mapping its parameter fields as inputs to the corresponding variables in the rule expression. The system then executes the rule chain verification process using a symbolic reasoning engine. This reasoning process, based on first-order logic or description logic semantics, determines whether the policy combination semantically satisfies all constraints. It also dynamically triggers the rule cascade along the dependency chain to ensure complete logical consistency within all rule sets. If a conflict, dependency violation, or precondition failure occurs during the reasoning process, the policy combination is immediately deemed non-compliant. In S1033, the system marks all policy combinations that have passed the symbolic rule chain verification as "compliant" and removes combinations that have not passed the verification from the candidate set.

[0028] Furthermore, the step S2 specifically includes the following sub-steps: S201. Generate a unique product entity code based on the coding rules, lifecycle, and incentive parameters in the strategy combination; and construct the corresponding digital twin code data structure based on the data content in the strategy combination; S202. Based on the unique product code, perform a dual-code binding operation on the physical code and the digital twin code, and synchronously write the bound code pair information into the distributed storage.

[0029] Furthermore, in step S201, the digital twin code data structure includes code scanning response logic, excitation control parameters and dynamic update identifiers.

[0030] Specifically, in S201, a unique identification code is generated for the product based on the coding rules contained in the selected strategy combination. The coding rules comprehensively consider the product life cycle management, incentive parameter settings and product attributes to ensure the uniqueness and traceability of the entity code in the entire link. At the same time, based on the detailed data content in the strategy combination, a corresponding digital twin code data structure is constructed. This data structure integrates the code scanning response logic, incentive control parameters and dynamic update identification, and can map the status and behavioral response of the product in the circulation process in real time to form a digital product mapping body. In S202, the generated unique code of the product is used as an anchor point, and the entity code and the digital twin code are bound one by one to establish a stable code pair mapping relationship. After the binding is completed, the relevant code pair information is synchronously written into the distributed storage environment to ensure efficient access and data consistency of multiple nodes.

[0031] Furthermore, the step S3 specifically includes the following sub-steps: S301. Based on the mapping relationship between the physical code and the digital twin code, collect the timestamp, user location, and response path of each product scan event to construct a time series of scan behavior. S302. Calculate the scan frequency, path deviation, and stimulus response rate based on the scan time series to form a feedback vector. Based on this feedback vector, invoke the Neural-Symbolic inference engine to analyze the adaptability and failure risk of the current strategy combination in actual behavior and determine whether to trigger strategy evolution. S303. When strategy evolution is triggered, a new bC strategy combination is dynamically generated based on the analysis results, and the response logic and incentive parameters in the corresponding digital twin code are updated in real time through the remote interface to complete strategy iteration. If strategy evolution is not triggered, step S301 is repeated.

[0032] Specifically, in S301, based on the mapping relationship between the physical code and the digital twin code, detailed data of each product scanning event is collected in real time, including the timestamp of the scan, the user's geographic location information, and the response link triggered by the scan. A complete scanning behavior time series is constructed through continuous events to reflect the dynamic interaction process of the product in the circulation link. In S302, the scanning time series is analyzed, the scanning frequency is calculated to measure user activity, the scanning path offset is measured to capture changes in geographical distribution, and the incentive response rate is statistically calculated to evaluate the actual effect of the incentive measures. These indicators are combined to form a multi-dimensional feedback vector. Subsequently, the feedback vector is input into the Neural-Symbolic reasoning engine. The system integrates the behavioral feature abstraction of the neural network and the rule constraints of symbolic logic to comprehensively evaluate the adaptability and potential failure risk of the current strategy combination in the real behavioral environment, and determine whether the threshold for triggering strategy evolution has been reached. In S303, if the trigger condition is met, a new bC strategy combination is dynamically generated based on the reasoning analysis results, including the updated b-end control strategy and C-end incentive path, and the corresponding response logic and incentive parameters in the digital twin code are synchronously updated through the remote interface to achieve real-time iterative optimization of the strategy; if the strategy evolution is not triggered, the code scanning data continues to be collected and S301 is executed in a loop to ensure that the strategy continues to be adaptively adjusted based on the latest behavior data.

[0033] Furthermore, step S302 specifically includes the following sub-steps: S3021. Extract the timestamp, spatial coordinates, and response link information from the scan event record to form a scan event set: , wherein the Represents a code scanning event set, Indicates the record of the i-th code scanning event, Respectively represent the timestamp of the i-th code scanning event, the horizontal coordinate of the spatial coordinate, the vertical coordinate of the spatial coordinate and the response link information, Indicates the number index of code scanning events. Indicates the total number of code scanning events collected within the statistical time range; S3022. Construct a scan interval vector based on the timestamp sequence corresponding to the timestamp: , wherein the Represents the code scanning interval vector, Indicates the time difference between the i-th scan and the i-1-th scan, which is used to reflect the time interval between scans. Indicates the timestamp of the i-1th scanning event; Based on the spatial coordinates, calculate the difference in the scan path vector: , wherein the Indicates the difference in the scanning path vector, which is used to reflect the change in the spatial distance between consecutive scanning events. represents the spatial distance between the i-th scanning event and the i-1-th scanning event, Represents the horizontal coordinate of the spatial coordinate of the i-1th scanning event, The vertical coordinate of the spatial coordinate of the i-1th scanning event; Combined with the response link strength sequence, a behavioral feature matrix is ​​formed: , wherein the Represents the behavioral feature matrix, a multidimensional feature set consisting of the interval between code scanning events, path differences, and response link information; S3023. Input the behavior feature matrix into a pre-trained neural network encoder, which outputs a feedback vector representing the low-dimensional features. Combined with the symbolic logic module of the policy rule library, the algorithm infers whether the current scanning behavior conforms to the expected policy and calculates the symbolic logic score. S3024. Define an evolution trigger function based on the neural network output and the symbolic logic score, and determine whether to trigger strategy evolution based on the evolution trigger function.

[0034] Furthermore, in step S3024, the evolution trigger function is specifically expressed as: , wherein the The strategy evolution trigger flag is a Boolean variable that indicates whether the strategy evolution is currently triggered. 1 indicates triggering, and 0 indicates not triggering. Represents the weight coefficient of the offset part of the behavior feedback vector, represents the feedback vector, represents the reference behavior feature vector, represents the weight of the symbolic logic module score s, represents the decision threshold that triggers evolution, that is, the sum of the maximum deviation and mismatch degree tolerated by the system; when =1, trigger strategy evolution and execute strategy combination adjustment; otherwise, maintain the current strategy.

[0035] Furthermore, the step S4 specifically includes the following sub-steps: S401. Based on the geographic coordinate data of the scan events recorded in the updated digital twin code, identify scan hotspots using a spatial clustering algorithm and calculate the scan density and activity level in each area. S402. Generate a heat map of the corresponding product's scan code distribution based on scan density and activity. S403. B-side staff use the warehouse management terminal to call the code scanning heat map, combine time and batch dimensions to analyze the code scanning active areas, and monitor product circulation and marketing strategies.

[0036] Specifically, in S401, the geographic coordinate data of the scanning events that are updated and recorded in real time in the digital twin code are used to preprocess the coordinate data, including removing outliers and standardizing the coordinates. A spatial clustering algorithm (such as DBSCAN or K-means) is then used to group the scanning points, identify hotspots with dense scanning, and calculate the scanning density and activity indicators based on the number and time distribution of scanning events in each area to quantify the scanning popularity and user interaction intensity in different regions. In S402, based on the scanning density and activity data of each hotspot area, combined with geographic information system technology, the data is converted into a visual scanning heat distribution map. This map intuitively reflects the user attention and scanning behavior strength of the product in different geographical locations, facilitating in-depth insights into regional market activity and consumption habits. In S403, the B-side staff accesses the code scanning heat map through the warehouse management end, and conducts a multi-dimensional analysis of the code scanning active areas in combination with the time dimension and batch information to identify sales trends and potential market opportunities, thereby scientifically adjusting and optimizing the commodity circulation path, inventory allocation and marketing strategy, and realizing precise management and decision support based on real-time data.

[0037] Example 2 Furthermore, as a preferred implementation of the above embodiment, a detailed calculation process example of the symbolic logic module described in step S3024 of the above embodiment is given. First, it is defined as follows: The bC strategy combination bound to the current product for: , each policy element Contains policy thresholds, execution conditions, and trigger logic structures; Anti-expensive vectors of code scanning behavior Expressed as: , respectively, the scanning frequency , frequency fluctuation , path deviation mean , path deviation fluctuation , stimulus response rate ; Each rule in the logic rule base is a formal logic expression: , indicating that a certain behavior feature should be mapped to a certain logical judgment of the strategy dimension. represents the kth symbolic logic rule, The premise of this rule is a logical expression of the feedback vector F, which determines whether the feedback of the code scanning behavior meets a certain condition. Indicates the conclusion of the rule, which is about the j-th strategy element in the current strategy combination A logical proposition indicating whether a policy element satisfies the rule requirements.

[0038] Secondly, construct the feedback logic mapping function , where each rule constructs a specific condition based on the feedback vector, for example: High-frequency scanning area: ; High path deviation fluctuations indicate discrete user activity: A low incentive response rate indicates a weak strategy appeal: ; These logical conditions all return Boolean values ; Among them, the 、 、 They represent the code scanning frequency threshold, path deviation threshold, and stimulus response rate threshold in the feedback logic, respectively, and are used to determine whether the feedback condition is triggered.

[0039] Again, construct the strategy proposition mapping function , for the current strategy combination A policy element in , the symbolic module extracts its logical proposition, for example: Strategy Proposition 1: Currently set the "weak incentive" parameters for the "low-frequency scanning area": Strategic Proposition 2: Currently, users with “dispersed paths” should not repeatedly receive the same type of incentive: ; Among them, the Indicates the excitation intensity threshold.

[0040] Again, the logic rule matching function Combine the feedback premise and the strategy proposition into an implicative logic: ; Exemplary: like and , then the rule does not match; like , logical implications are always true (no rule violation); final .

[0041] Finally, the symbolic logic scores are summarized Perform weighted summation on all rules to form the final symbolic logic adaptation score: ; Among them, the The index representing the number of all symbolic logic rules in the rule base, Represents the number of all symbolic logic rules in the rule base, It represents the weight of the k-th symbolic logic rule, that is, the importance of the rule in the overall judgment.

[0042] Example 3 Furthermore, as a preferred implementation of the above embodiment, an application scenario of a bC integrated marketing method based on one item, one code is proposed. Specifically, a large fast-moving consumer goods company wants to improve the market penetration and user engagement of its newly launched beverage products. The bC integrated marketing method based on one item, one code of the present invention is adopted, combined with Neural-Symbolic reasoning technology, to achieve dynamic strategy adjustment and precise incentives for the entire chain of commodity circulation.

[0043] The company first collects data on beverage product attributes, target sales regions, and user tags. Through structured data processing, a standardized feature vector is constructed and input into the Neural-Symbolic reasoning framework. The neural network model matches candidate business-to-consumer (BC) strategy combinations with historically strong performance, such as "spotting scheduling - scan code to receive coupons" and "shelf control - behavioral chain guidance." The symbolic reasoning engine selects compliant strategies based on company-defined rules (inventory limits, campaign compliance), and combines multi-objective optimization to determine the optimal strategy combination. For example, spotting scheduling prioritizes East China, while the consumer-end incentive path utilizes scan code rewards and sharing fission.

[0044] Based on the determined strategy combination, the system automatically generates a unique physical code, appending the batch number and lifecycle information, which is printed on the beverage bottle. Simultaneously, a digital twin code is generated, containing the points reward rules, point amount, expiration date, and dynamic update identifier. This digital twin is then bound to the physical code through the product's unique identifier and written into a distributed database.

[0045] After the product is launched, the system collects scanning event data in real time, including the scanning timestamp, geographic location, and the response link after the user scans the code. Based on the scanning time series, the system calculates the scanning frequency, path deviation, and incentive response rate to form a feedback vector. Input the Neural-Symbolic reasoning engine to calculate the symbolic logic score and evaluate the adaptability of the current strategy. When the scanning behavior deviates from expectations, that is, the scanning rate of a store decreases, the sharing fission is insufficient, and the symbolic logic score is lower than the threshold, the strategy evolution is triggered. The system automatically generates a new bC strategy combination, such as increasing the amount of scanning reward points or adjusting the sharing incentive method, and updates the response logic of the digital twin code in real time through the remote interface to ensure a rapid closed loop between strategy and market feedback.

[0046] Based on the geographic coordinates of scan events recorded by the digital twin code, spatial clustering algorithms such as DBSCAN are used to identify scanning hotspots and calculate scanning density and activity. A scanning heat map is generated, noting active stores and low-volume areas. Warehouse managers on the business side can view this heat map through the management terminal and analyze product distribution status by combining batch and time dimensions. This data-driven approach allows them to adjust warehouse allocation and marketing strategies, increasing promotional resources in areas with low scanning density as indicated by the heat map.

[0047] The foregoing description is merely a preferred embodiment of the present invention. It should be understood that the present invention is not limited to the form disclosed herein and should not be construed as excluding other embodiments. Rather, the present invention can be used in various other combinations, modifications, and environments and can be modified within the scope of the concept described herein through the above teachings or techniques or knowledge in the relevant field. Modifications and variations made by those skilled in the art that do not depart from the spirit and scope of the present invention are intended to be protected by the appended claims.

Claims

1. A bC integrated marketing method based on one item one code, characterized by: The following steps are involved: S1. Generate a feature vector for the product based on its attribute parameters, sales scenario, and user tags. Combined with Neural-Symbolic methods, the system infers the b / c strategy combination bound to the feature vector. Jointly determine the b / c control strategy and c / c incentive path to be adopted for the product at the current stage, and output the b / c strategy combination structure. S2. Based on the inferred bC strategy combination, a physical code is generated for the product. A digital twin code containing the response logic is also generated for each physical code. The physical code and digital twin code are bound together using the product's unique code, forming a one-to-one code pair mapping relationship. S3. Collect the timestamp, spatial coordinates, and response link information from the product scanning behavior event and input it into the Neural-Symbolic structure for behavior analysis and policy adjustment. The new policy combination is output and the logical content of the original digital twin code is updated based on the new policy combination. S4. Based on the scan events recorded in the updated digital twin code, coordinate clustering statistics are performed on the C-end scan location field to generate a scan density distribution map. The thermal distribution results are then called by product number on the warehouse side to visualize the B-end warehouse.

2. The bC integrated marketing method based on one item one code according to claim 1, characterized in that: The step S1 specifically includes the following sub-steps: S101. Structuralize product attribute data based on product category, SKU, target sales region, and user profile, construct a standardized feature vector, and input this standardized feature vector into the Neural-Symbolic fusion inference framework. S102. Calling the trained neural network model to perform similarity matching between the performance characteristics of the product and the bC strategy combinations in the historical database to identify potential effective strategy combination candidates, wherein the bC strategy combination includes the b-side control strategy and the c-side incentive path; S103. Based on the policy constraints set in the business rule library, the symbolic reasoning engine performs compliance screening and logical verification on the candidate policy combinations output by the neural network, eliminating policy paths that do not comply with the rules; S104. Based on the strategy set filtered by symbolic reasoning, a multi-objective optimization method is used for joint decision-making to determine and output the optimal strategy combination of the B-end behavior control strategy and the C-end incentive path that should be bound to the product at the current stage.

3. The bC integrated marketing method based on one item one code as claimed in claim 2, characterized in that: In step S102, the B-end control strategy includes warehouse scheduling, shelf layout control, inventory rhythm, batch strategy, code life cycle setting and product permission configuration, and the C-end incentive path includes code scanning reward parameters, behavior chain guidance, incentive presentation form, user verification mechanism, cumulative behavior tracking, sharing fission path and incentive dynamic update mechanism.

4. The bC integrated marketing method based on one item one code according to claim 2, characterized in that: The step S102 specifically includes the following sub-steps: S1021 reads the stored strategy combination vector from the historical bC strategy database, each strategy combination vector corresponds to the performance characteristics of a strategy combination; S1022. Calculate the weighted cosine similarity between the normalized feature vector of the current product and each historical strategy combination vector; S1023. Using a preset similarity threshold as a screening criterion, select the strategy combinations that are most relevant to the current product and have good historical performance to form a candidate strategy set.

5. The bC integrated marketing method based on one item one code according to claim 2, characterized in that: The step S103 specifically includes the following sub-steps: S1031. Load the current business rule library involving constraints on the B-side control and C-side incentive path; S1032. For each candidate strategy combination, use the symbolic reasoning engine to apply business rules in sequence, map the strategy combination parameters to the rule expression and perform logical verification; S1033. Mark the policy combinations that fully meet all rule conditions as compliant, eliminate the non-compliant policy combinations, and output the policy set after symbolic reasoning screening.

6. The bC integrated marketing method based on one item one code according to claim 1, characterized in that: The step S2 specifically includes the following sub-steps: S201. Generate a unique product entity code based on the coding rules, lifecycle, and incentive parameters in the strategy combination; and construct the corresponding digital twin code data structure based on the data content in the strategy combination; S202. Based on the unique product code, perform a dual-code binding operation on the physical code and the digital twin code, and synchronously write the bound code pair information into the distributed storage.

7. The bC integrated marketing method based on one item one code according to claim 6, characterized in that: In step S201, the digital twin code data structure includes code scanning response logic, excitation control parameters and dynamic update identifier.

8. The bC integrated marketing method based on one item one code according to claim 1, characterized in that: The step S3 specifically includes the following sub-steps: S301. Based on the mapping relationship between the physical code and the digital twin code, collect the timestamp, user location, and response path of each product scan event to construct a time series of scan behavior. S302. Calculate the scan frequency, path deviation, and stimulus response rate based on the scan time series to form a feedback vector. Based on this feedback vector, invoke the Neural-Symbolic inference engine to analyze the adaptability and failure risk of the current strategy combination in actual behavior and determine whether to trigger strategy evolution. S303. When strategy evolution is triggered, a new bC strategy combination is dynamically generated based on the analysis results, and the response logic and incentive parameters in the corresponding digital twin code are updated in real time through the remote interface to complete strategy iteration. If strategy evolution is not triggered, step S301 is repeated.

9. The bC integrated marketing method based on one item one code according to claim 8, characterized in that: The step S302 specifically includes the following sub-steps: S3021. Extract the timestamp, spatial coordinates, and response link information from the scan event record to form a scan event set: ; S3022. Construct a scan interval vector based on the timestamp sequence corresponding to the timestamp: , wherein the Represents the code scanning interval vector, Indicates the time difference between the i-th scan and the i-1-th scan, which is used to reflect the time interval between scans. Indicates the timestamp of the i-1th scanning event; Based on the spatial coordinates, calculate the difference in the scan path vector: , wherein the Indicates the difference in the scanning path vector, which is used to reflect the change in the spatial distance between consecutive scanning events. represents the spatial distance between the i-th scanning event and the i-1-th scanning event, Represents the horizontal coordinate of the spatial coordinate of the i-1th scanning event, The vertical coordinate of the spatial coordinate of the i-1th scanning event; Combined with the response link strength sequence, a behavioral feature matrix is ​​formed: , wherein the Represents the behavioral feature matrix, a multidimensional feature set consisting of the interval between code scanning events, path differences, and response link information; S3023. Input the behavior feature matrix into a pre-trained neural network encoder, which outputs a feedback vector representing the low-dimensional features. Combined with the symbolic logic module of the policy rule library, the algorithm infers whether the current scanning behavior conforms to the expected policy and calculates the symbolic logic score. S3024. Define an evolution trigger function based on the neural network output and the symbolic logic score, and determine whether to trigger strategy evolution based on the evolution trigger function.

10. The bC integrated marketing method based on one item one code according to claim 9, characterized in that: In step S3024, the evolution trigger function is specifically expressed as: , wherein the The strategy evolution trigger flag is a Boolean variable that indicates whether the strategy evolution is currently triggered. 1 indicates triggering, and 0 indicates not triggering. Represents the weight coefficient of the offset part of the behavior feedback vector, represents the feedback vector, represents the reference behavior feature vector, represents the weight of the symbolic logic module score s, represents the decision threshold that triggers evolution, that is, the sum of the maximum deviation and mismatch degree tolerated by the system; when =1, trigger strategy evolution and execute strategy combination adjustment; otherwise, maintain the current strategy.

11. The bC integrated marketing method based on one item one code according to claim 1, characterized in that: The step S4 specifically includes the following sub-steps: S401. Based on the geographic coordinate data of the scan events recorded in the updated digital twin code, identify scan hotspots using a spatial clustering algorithm and calculate the scan density and activity level in each area. S402. Generate a heat map of the corresponding product's scan code distribution based on scan density and activity. S403. B-side staff use the warehouse management terminal to call the code scanning heat map, combine time and batch dimensions to analyze the code scanning active areas, and monitor product circulation and marketing strategies.

Images