Big data-based precision marketing customer positioning method, device, equipment and medium

Abstract

This application relates to the field of data processing technology. It provides a method, apparatus, device, and medium for precise marketing customer positioning based on big data. The method includes: integrating heterogeneous data from a customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset; based on the panoramic customer profile dataset, performing customer segmentation processing through a pre-set multi-level clustering engine to generate a positioning coding matrix; performing strategy knowledge base mapping and matching processing on the positioning coding matrix to generate a differentiated strategy instruction set; and distributing the differentiated strategy instruction set across all channels to the customer touchpoint system for execution. This approach overcomes the barriers of multi-source data formats, integrates behavioral characteristics and demand evolution patterns, and enhances the scenario adaptability and terminal compatibility of strategy distribution.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method, apparatus, equipment and medium for precise marketing customer positioning based on big data. Background Technology

[0002] With the widespread application of big data technology, precision marketing has become a core means for enterprises to enhance their market competitiveness. Customer positioning, as a key component of precision marketing, aims to achieve accurate identification and differentiated operations for customers of varying value through multi-dimensional data analysis. Efficient customer positioning technology can significantly improve the utilization rate of marketing resources, optimize customer experience, and drive business growth.

[0003] Traditional technologies use conventional format adaptation methods to integrate multi-type customer data, but this is difficult to overcome the heterogeneous barriers of structured, time-series, and graph data, hindering the construction of a comprehensive customer view. Customer segmentation relies on traditional clustering techniques that only focus on static attributes and lack the ability to capture the spatiotemporal characteristics of behavior and the evolution of demand patterns, resulting in segmentation results that are out of touch with business realities. In addition, policy distribution relies on fixed manual rule bases and protocol conversion methods, lacking real-time scenario adaptation and terminal context awareness capabilities, resulting in poor policy execution efficiency. Summary of the Invention

[0004] Therefore, it is necessary to provide big data-based precise marketing customer positioning methods, devices, equipment, and media to address the aforementioned technical issues, in order to break through the barriers of multi-source data formats, integrate behavioral characteristics and demand evolution patterns, and enhance the scenario adaptability and terminal compatibility of strategy distribution.

[0005] Firstly, this application provides a method for precise marketing customer targeting based on big data, which includes:

[0006] Heterogeneous data integration and processing are performed on the customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset.

[0007] Based on the panoramic customer profile dataset, a pre-set multi-level clustering engine is used to perform customer segmentation processing and generate a location coding matrix.

[0008] The positioning encoding matrix is ​​mapped and matched using a strategy knowledge base to generate a differentiated strategy instruction set;

[0009] The differentiated strategy instruction set is distributed across all channels and output to customer touchpoint systems for execution.

[0010] In one embodiment, based on a panoramic customer profile dataset, a pre-defined multi-level clustering engine is used to perform customer segmentation processing to generate a location coding matrix, including:

[0011] The following formula is used to perform joint clustering of contribution and activity levels on the panoramic customer profile dataset to generate value-level grouping sets:

[0012]

[0013] in, Represents a value hierarchy grouping set. Indicates the first Each customer's contribution feature vector Indicates the first The contribution cluster centers of the group Indicates the first Customer activity feature vector Indicates the first The activity cluster centers of the group This represents the contribution weighting coefficient. This represents the minimization optimization operator. express A set of groups, Indicates the first A collection of customers at different value levels;

[0014] Based on the value hierarchy grouping set, deep learning processing of interaction trajectories is performed within each value hierarchy group to generate behavioral feature sub-clusters;

[0015] Perform response pattern correlation analysis on behavioral feature subclusters to generate a set of demand dimension labels;

[0016] The value tags of the value hierarchy group set, the behavior tags of the behavior feature subgroup, and the demand tags of the demand dimension tag set are combined and encoded to generate a positioning encoding matrix.

[0017] In one embodiment, based on a value hierarchy grouping set, deep learning processing of interaction trajectories is performed within each value hierarchy group to generate behavioral feature sub-clusters, including:

[0018] Multi-dimensional feature decoupling processing is performed on user interaction trajectories within the value hierarchy group set to generate a spatiotemporal sequence feature set.

[0019] Deep pattern extraction is performed on the spatiotemporal sequence feature set using an attention-enhanced recurrent neural network to generate a set of behavioral prototype vectors.

[0020] Dynamic density clustering is performed based on semantic similarity in the behavioral prototype vector set to generate behavioral feature sub-clusters.

[0021] Interactive intent reasoning is performed on the behavioral feature subgroups to generate behavioral feature subgroups with semantic labels.

[0022] In one embodiment, the differentiated strategy instruction set is distributed across all channels and output to the customer touchpoint system for execution, including:

[0023] Perform channel compatibility analysis on the differentiated strategy instruction set to generate an adapted distribution path set;

[0024] Based on the adapted distribution path set, the policy instructions are processed by protocol conversion and format encapsulation to generate a channel executable instruction set;

[0025] The executable instruction set of the channel is load-balanced and scheduled through an asynchronous message bus, and then output to the customer touchpoint system for execution.

[0026] Collect execution feedback data from customer touchpoint systems to generate quantitative parameters for distribution effectiveness.

[0027] In one embodiment, based on the adapted distribution path set, the policy instructions are subjected to protocol conversion and format encapsulation processing to generate a channel-executable instruction set, including:

[0028] Perform channel protocol feature parsing on the adapted distribution path set to generate a protocol feature descriptor set;

[0029] Based on the protocol feature descriptor set, an intermediate protocol instruction set is generated by performing instruction syntax conversion processing through a preset dynamic protocol conversion rule base.

[0030] The intermediate protocol instruction set is adaptively restructured by combining user terminal context data to generate a channel executable instruction set.

[0031] The channel executable instruction set is subjected to fault-tolerant redundancy coding processing to generate a channel executable instruction set with a self-verification mechanism.

[0032] In one embodiment, the positioning encoding matrix is ​​subjected to policy knowledge base mapping and matching processing to generate a differentiated policy instruction set, including:

[0033] The positioning encoding matrix is ​​subjected to hierarchical bit sequence decoding to generate a set of value level identifiers, a set of behavior pattern identifiers, and a set of demand type identifiers.

[0034] Based on the preset triple association rule base, the value level identifier set, behavior pattern identifier set and demand type identifier set are jointly retrieved and processed to generate an initial strategy matching set.

[0035] The initial policy matching set is dynamically optimized by integrating real-time scene parameters to generate an enhanced policy instruction set;

[0036] Perform policy conflict detection and compliance filtering on the enhanced policy instruction set to generate a differentiated policy instruction set.

[0037] In one embodiment, the distribution effectiveness quantification parameters are obtained using the following formula:

[0038]

[0039] in, This represents the quantification parameter of distribution effectiveness. This represents the efficiency quantification function. Indicates the number of customer touchpoint systems. Represents the message bus state matrix. This represents the load balancing scheduling function. This represents protocol conversion and format encapsulation functions. Indicates the set of distribution paths to be adapted. This represents the instruction set for differentiated strategies.

[0040] Secondly, this application also provides a big data-based precision marketing customer targeting device, which includes:

[0041] The profile integration module is used to integrate heterogeneous data from the customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset.

[0042] The clustering coding module is used to perform customer segmentation processing based on the panoramic customer profile dataset and generate a location coding matrix through a preset multi-level clustering engine.

[0043] The strategy mapping module is used to perform strategy knowledge base mapping and matching on the positioning encoding matrix to generate a differentiated strategy instruction set;

[0044] The channel distribution module is used to distribute differentiated strategy instruction sets across all channels and output them to customer touchpoint systems for execution.

[0045] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of any of the methods in the first aspect of this application.

[0046] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of any of the methods in the first aspect of this application.

[0047] The present application provides a method, apparatus, equipment and medium for precise marketing customer positioning based on big data. The method includes heterogeneous data integration processing of customer transaction database, behavior log database and attribute knowledge base. It can incorporate structured transaction data, time-series behavior log and graph structure attribute knowledge into a unified processing framework. By breaking down the format barriers between different data sources, it can construct a panoramic customer profile dataset containing multi-dimensional information, thereby achieving a breakthrough in the multi-source data format barriers.

[0048] Based on a pre-defined multi-level clustering engine using a panoramic customer profile dataset, this approach not only incorporates static attributes during customer segmentation but also deeply mines dynamic features within behavioral trajectories. It combines interaction pattern analysis to capture the evolutionary patterns of needs over time. Through combined encoding of value levels, behavioral subgroups, and demand tags, it achieves the fusion of behavioral characteristics and demand evolution patterns. For the omnichannel distribution of differentiated strategy instruction sets, it determines adaptation paths through channel compatibility analysis, performs protocol conversion and format encapsulation based on terminal context, and achieves load balancing scheduling through an asynchronous message bus. This effectively adapts to the characteristics of terminals in different scenarios, enhancing the scenario adaptability and terminal compatibility of strategy distribution. Attached Figure Description

[0049] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1 This is a flowchart of a big data-based precision marketing customer positioning method according to one embodiment of the present invention;

[0051] Figure 2 This is a flowchart illustrating how, in one embodiment of the present invention, deep learning processing of interaction trajectories is performed within each value-level group to generate behavioral feature subclusters.

[0052] Figure 3 This is a structural diagram of a big data-based precision marketing customer positioning device according to one embodiment of the present invention. Detailed Implementation

[0053] To make the above-mentioned objects, features, and advantages of this application more apparent and understandable, the specific embodiments of this application will be described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of the application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0054] First, the application scenarios of the embodiments of this application are described. In the embodiments of this application, a method, apparatus, device, and medium for precise marketing customer positioning based on big data are provided, applicable to scenarios such as precision marketing promotion on e-commerce platforms, differentiated customer services for financial institutions, matching user needs of retail enterprises, and personalized operation of internet service platforms.

[0055] The big data-based precision marketing customer positioning method, device, equipment, and medium provided in this application embodiment can also be applied to other application scenarios that require precise reach through customer positioning, such as target customer mining in the cultural and tourism industry, potential user conversion in educational institutions, and precise push of medical and health services. This is only an example and does not limit the specific application scenarios.

[0056] like Figure 1 As shown, this application provides a method for precise marketing customer targeting based on big data, the method including:

[0057] S101: Perform heterogeneous data integration processing on the customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset.

[0058] For example, data cleaning and standardization processes are performed on the original transaction data in the customer transaction database, the original behavior record data in the behavior log database, and the original customer attribute data in the attribute knowledge base to remove invalid information and redundant content, and to unify data description standards and format specifications.

[0059] Based on preset data association rules, cross-source association matching is performed on the standardized customer transaction database data, behavior log database data, and attribute knowledge base data to establish logical mapping relationships between the various data sources and eliminate heterogeneous differences between the multi-source data. Data fusion technology is then used to integrate and summarize the associated multi-dimensional data, forming a complete dataset including customer transaction characteristics, behavioral characteristics, and attribute characteristics, generating a panoramic customer profile dataset.

[0060] S102: Based on the panoramic customer profile dataset, customer segmentation is performed through a preset multi-level clustering engine to generate a location coding matrix.

[0061] For example, customer value, behavior, and demand-related features are extracted from a panoramic customer profile dataset and input into a preset multi-level clustering engine. The first level of clustering forms customer value hierarchical groups, and within each group, the second level of clustering further divides customer behavior feature subgroups. Customer demand dimension tags are then mined, and the value tags, behavior tags, and demand tags are standardized and combined into a localization coding matrix.

[0062] S103: Perform policy knowledge base mapping and matching on the positioning encoding matrix to generate a differentiated policy instruction set.

[0063] For example, the location encoding matrix is ​​decoded to extract customer value identifiers, behavioral pattern identifiers, and demand type identifiers. According to preset association rules, these identifiers are mapped and matched with policy entries in the policy knowledge base to filter and adapt an initial policy set. The set is then optimized by integrating real-time scenario parameters and business compliance requirements, eliminating conflicting and non-compliant content. The optimized policies are then encapsulated into standardized instructions, clearly defining execution dimensions, triggering conditions, and applicable scenarios to generate a differentiated policy instruction set.

[0064] S104: Performs omnichannel distribution processing on the differentiated strategy instruction set and outputs it to the customer touchpoint system for execution.

[0065] For example, a compatibility analysis is performed on the differentiated strategy instruction set with omnichannel technical characteristics and terminal adaptation requirements to clarify the adaptation and distribution path. Combining channel protocol specifications and terminal context, the differentiated strategy instruction set is protocol-converted and standardized to generate channel-executable instructions. These channel-executable instructions are distributed through a load balancing scheduling mechanism to balance node pressure. The scheduled channel-executable instructions are then output to the customer touchpoint system for execution.

[0066] An embodiment of this application provides a big data-based precision marketing customer positioning method, which includes heterogeneous data integration processing of customer transaction database, behavior log database and attribute knowledge base. It can incorporate structured transaction data, time-series behavior logs and graph structure attribute knowledge into a unified processing framework. By breaking down the format barriers between different data sources, it can construct a panoramic customer profile dataset containing multi-dimensional information, thereby achieving a breakthrough in the multi-source data format barriers.

[0067] Based on a pre-defined multi-level clustering engine using a panoramic customer profile dataset, this approach not only incorporates static attributes during customer segmentation but also deeply mines dynamic features within behavioral trajectories. It combines interaction pattern analysis to capture the evolutionary patterns of needs over time. Through combined encoding of value levels, behavioral subgroups, and demand tags, it achieves the fusion of behavioral characteristics and demand evolution patterns. For the omnichannel distribution of differentiated strategy instruction sets, it determines adaptation paths through channel compatibility analysis, performs protocol conversion and format encapsulation based on terminal context, and achieves load balancing scheduling through an asynchronous message bus. This effectively adapts to the characteristics of terminals in different scenarios, enhancing the scenario adaptability and terminal compatibility of strategy distribution.

[0068] In one embodiment, based on a panoramic customer profile dataset, a pre-defined multi-level clustering engine is used to perform customer segmentation processing to generate a location coding matrix, including:

[0069] (1) Using the following formula, perform contribution-activity joint clustering on the panoramic customer profile dataset to generate a value-level grouping set:

[0070]

[0071] in, Represents a value hierarchy grouping set. Indicates the first Each customer's contribution feature vector Indicates the first The contribution cluster centers of the group Indicates the first Customer activity feature vector Indicates the first The activity cluster centers of the group This represents the contribution weighting coefficient. This represents the minimization optimization operator. express A set of groups, Indicates the first A collection of customers at different value levels.

[0072] For example, based on a panoramic customer profile dataset, feature indicators related to customer contribution and those related to customer activity are selected, and these indicators are transformed into structured contribution feature vectors and activity feature vectors. According to preset contribution weight coefficients, the contribution feature vectors and activity feature vectors are weighted and fused. By calculating the distance between each customer feature vector and potential cluster centers, a minimization optimization operator is used to iteratively adjust the cluster center positions and customer group affiliations until the feature differences within each group are minimized, thus dividing customers into multiple value levels and generating value level group sets.

[0073] The value tier grouping set contains multiple customer sets, each consisting of customers with similar value characteristics, corresponding to a specific value tier.

[0074] (2) Based on the value hierarchy grouping set, deep learning processing of interaction trajectory is performed in each value hierarchy grouping to generate behavioral feature sub-clusters.

[0075] For example, for each value-level group, customer interaction trajectory data under different scenarios is extracted within each value-level group. The trajectory data is cleaned to remove abnormal records and aligned according to time series. The processed interaction trajectory data is input into an attention-enhanced deep learning model to extract spatiotemporal correlation features and behavioral pattern features from the trajectories, generating behavioral prototype vectors. The semantic similarity between each behavioral prototype vector is calculated, and cluster boundaries are dynamically divided based on a similarity threshold. Customers with similar behavioral patterns are grouped into the same subgroup, generating behavioral feature subclusters.

[0076] Among them, the behavioral feature sub-clusters are further subdivided within each value level group, and each sub-cluster contains customers with similar behavioral patterns.

[0077] (3) Perform response pattern association analysis on the behavioral feature sub-clusters to generate a set of demand dimension labels.

[0078] For example, for each subgroup of the behavioral characteristic subgroup, customer response data to various marketing stimuli in historical interactions is collected, and response pattern features such as frequency, timeliness, and preferences are extracted. The mapping relationship between response pattern features and potential customer needs is mined using association analysis methods, and a classification standard for demand dimensions is defined in conjunction with business scenarios. Based on the mapping relationship and classification standard, each subgroup is assigned a corresponding demand dimension identifier, generating a demand dimension label set.

[0079] The demand dimension tag set is a collection of tags that identify the types of customer needs, with each tag corresponding to a subgroup of customers with common needs.

[0080] (4) Combine and encode the value tags of the value hierarchy group set, the behavior tags of the behavior feature subgroup, and the demand tags of the demand dimension tag set to generate a positioning encoding matrix.

[0081] For example, value tags corresponding to each value level are extracted from the value level grouping set, behavior tags corresponding to each subgroup are extracted from the behavior feature subgroup, and demand tags corresponding to each subgroup are extracted from the demand dimension tag set. According to preset coding rules, the value tags, behavior tags, and demand tags corresponding to the same customer group are hierarchically combined and transformed into a structured coding sequence. The coding sequences of all customer groups are arranged according to a unified dimension to form a matrix structure, generating a positioning coding matrix.

[0082] The location coding matrix is ​​a matrix containing three types of tag codes: customer value, behavior, and needs. Each matrix element corresponds to the comprehensive location information of a specific customer group.

[0083] like Figure 2 As shown, based on the value hierarchy grouping set, deep learning processing of interaction trajectories is performed within each value hierarchy group to generate behavioral feature sub-clusters, including:

[0084] S201: Perform multi-dimensional feature decoupling processing on user interaction trajectories within the value hierarchy group set to generate a spatiotemporal sequence feature set.

[0085] For example, for user interaction trajectories within a value-level group set, features such as the time of interaction occurrence, duration, and interval are extracted in the time dimension, while features such as interaction channels, page locations, and functional modules are extracted in the spatial dimension. Outlier filtering and missing value completion are performed on the extracted time and spatial features. A standardization method is used to unify the feature dimensions, and the processed time and spatial features are transformed into ordered numerical sequences. These two types of sequences are then integrated to form a feature set, generating a spatiotemporal sequence feature set.

[0086] Among them, the spatiotemporal sequence feature set is a collection of time-dimensional feature sequences and spatial-dimensional feature sequences containing user interaction trajectories, which fully preserves the spatiotemporal attributes of the interaction.

[0087] S202: Deep pattern extraction processing of spatiotemporal sequence feature sets is performed through attention-enhanced recurrent neural networks to generate behavioral prototype vector sets.

[0088] For example, the spatiotemporal sequence feature set is segmented into segments of fixed length and input into the input layer of an attention-enhanced recurrent neural network. The recurrent neural network captures the temporal dependencies within the sequence through the recurrent structure of the hidden layers, generating preliminary feature representations. The attention mechanism calculates weight coefficients for the preliminary feature representations of each segment, strengthening the influence of key segment features and weakening secondary segment features. The weighted and fused features are then compressed in dimension and have their core information extracted by a fully connected layer, outputting feature vectors representing behavioral patterns. All vectors are then aggregated to form a set, generating a behavioral prototype vector set.

[0089] Among them, the behavioral prototype vector set consists of core feature vectors extracted through deep learning that represent typical user behavior patterns, and the differences between the vectors reflect the different behavior patterns.

[0090] S203: Dynamic density clustering is performed based on semantic similarity in the behavioral prototype vector set to generate behavioral feature sub-clusters.

[0091] For example, the semantic similarity between any two vectors in the behavioral prototype vector set is calculated, and a similarity matrix is ​​generated using a preset algorithm. Based on the similarity matrix, a dynamic density threshold and neighborhood radius are set, and core vectors with densities exceeding the threshold are identified. The core vectors and all similar vectors within their neighborhoods that meet the similarity threshold are grouped into the same cluster. Non-core vectors are grouped according to their nearest core vector. This clustering process is repeated until all behavioral prototype vectors are assigned to a specific cluster. Each cluster corresponds to a group of customers with similar behavioral patterns. These customer groups are integrated to form a set, generating a behavioral feature sub-cluster.

[0092] Among them, the behavioral feature subgroup is a set of multiple customer subgroups, and customers in each subgroup have similar behavioral prototype vectors, that is, similar behavioral patterns.

[0093] S204: Perform interactive intent reasoning processing on the behavioral feature subgroups to generate behavioral feature subgroups with semantic labels.

[0094] For example, interaction content, interaction paths, and interaction result data of customers within each behavioral feature subgroup are collected. Combined with the interaction intent classification system of the business domain, an intent reasoning rule base is constructed. The interaction data of the subgroups are matched with the rule base to identify frequently occurring interaction goals and behavioral tendencies. Based on the matching results, the dominant interaction intent of each subgroup is determined, and corresponding semantic labels are assigned to the subgroups according to preset labeling specifications. The labeled subgroups are integrated to form a set, generating a behavioral feature subgroup with semantic labels.

[0095] Among them, the behavioral feature sub-clusters with semantic labels are based on the behavioral feature sub-clusters, and each sub-cluster has a set of interaction intent semantic identifiers added. The semantic labels intuitively reflect the behavioral purpose of the sub-clusters.

[0096] In one embodiment, the differentiated strategy instruction set is distributed across all channels and output to the customer touchpoint system for execution, including:

[0097] (1) Perform channel compatibility analysis on the differentiated strategy instruction set to generate an adapted distribution path set.

[0098] For example, for the differentiated strategy instruction set, the technical specifications of each marketing channel are analyzed, including communication protocol version, data transmission rate limits, and terminal system compatibility range. Execution parameters of the differentiated strategy instruction set are extracted, including characteristics such as instruction data volume, interaction response time requirements, and content format type. The channel technical specifications are matched with the instruction execution parameters in multiple dimensions. A list of channels that can stably transmit each strategy instruction is identified through compatibility verification rules, and the transmission priority and path nodes of each channel are marked. Based on the channel list and marking information, a specific distribution route for each instruction is planned, and all routes are integrated into a structured set to generate an adapted distribution path set.

[0099] The adapted distribution path set is a collection of channel information, transmission routes, and priorities for each differentiated strategy instruction, providing a path basis for subsequent distribution.

[0100] (2) Based on the adapted distribution path set, the policy instructions are processed by protocol conversion and format encapsulation to generate a channel executable instruction set.

[0101] For example, based on the adapted distribution path set, the channel protocol details corresponding to each path are parsed, including data frame structure, field naming conventions, and verification mechanisms. For each instruction in the differentiated strategy instruction set, the data organization of the instruction is adjusted according to the protocol details, and field names and data types are converted to match protocol requirements. Simultaneously, contextual data such as the user terminal's hardware model, software version, and screen size are collected, and the converted instructions are format-adapted according to terminal characteristics, such as adjusting content display layout and compressing data volume. Integrity verification is performed on the instructions that have completed protocol conversion and format adaptation to ensure no data loss or format errors. The verified instructions are then integrated into a set to generate a channel-executable instruction set.

[0102] Among them, the channel executable instruction set is a set of strategy instructions that can be directly parsed and executed by various marketing channels after protocol conversion and terminal format adaptation.

[0103] (3) The channel executable instruction set is load balanced and scheduled through the asynchronous message bus and output to the customer touch point system for execution.

[0104] For example, the channel-executable instruction set is imported into the task queue of the asynchronous message bus. The bus starts a real-time monitoring unit to continuously collect the current load data of each customer touchpoint system, including CPU utilization, memory usage, and the length of the pending task queue. Based on the load data, a dynamic load balancing algorithm is used to calculate the remaining processing capacity of each system and allocate tasks from the channel-executable instruction set according to the capacity ratio, avoiding overload of a single system. After allocation, the bus pushes the instructions to the corresponding customer touchpoint system via asynchronous communication. Upon receiving the instructions, the system triggers the execution process to complete the implementation of the policy instructions.

[0105] Among them, the asynchronous message bus is a middleware that supports asynchronous data transmission and has load monitoring and task allocation functions. It is used to coordinate the efficient distribution of channel executable instructions to customer touchpoint systems.

[0106] (4) Collect execution feedback data from the customer touchpoint system and generate distribution efficiency quantification parameters.

[0107] For example, in the entire process of executing executable instructions through the customer touchpoint system, data acquisition nodes are deployed to record data at each stage, including the time the instruction is issued from the bus, the system's receipt and confirmation time, the instruction parsing time, the execution result status, and the customer's interaction feedback content. The collected raw feedback data is preprocessed to remove duplicate records, correct outliers, and complete missing fields. The preprocessed data is then input into a distribution efficiency quantification model to calculate efficiency indicators such as instruction reach rate, execution latency, and customer response conversion rate. These indicators are standardized and integrated to form a quantifiable parameter set, generating distribution efficiency quantification parameters.

[0108] Among them, the distribution efficiency quantification parameters are a set of indicators obtained by analyzing and calculating execution feedback data, which are used to measure the execution efficiency and effectiveness of strategy instructions in the omnichannel distribution process.

[0109] In one embodiment, based on the adapted distribution path set, the policy instructions are subjected to protocol conversion and format encapsulation processing to generate a channel-executable instruction set, including:

[0110] (1) Perform channel protocol feature parsing on the adapted distribution path set to generate a protocol feature descriptor set.

[0111] For example, for the set of adapted distribution paths, the channel identification information contained therein is sorted out to clarify the target marketing channel corresponding to each distribution path. For each target marketing channel, the core elements of the channel protocol are extracted, including features such as protocol type, data frame format, field definition specifications, transmission verification rules, and data encoding methods. The extracted protocol features are structured and organized to form protocol feature description information for each channel according to a unified description specification. All description information is integrated to form a set, generating a protocol feature descriptor set.

[0112] Among them, the protocol feature descriptor set is a collection of structured descriptions of the protocol features of each channel in the adaptation and distribution path set, providing a clear basis for subsequent protocol conversion.

[0113] (2) Based on the protocol feature descriptor set, the instruction syntax conversion is performed through the preset dynamic protocol conversion rule base to generate an intermediate protocol instruction set.

[0114] For example, based on the protocol feature descriptor set, a preset dynamic protocol conversion rule library is invoked. This rule library contains syntax conversion rules, field mapping relationships, data format adaptation logic, and other content corresponding to different protocol types. For each instruction content of the strategy command, the corresponding channel protocol feature in the protocol feature descriptor set is used to match the appropriate conversion rule from the rule library. The syntax structure of the strategy command is adjusted according to the matched rule, and the field names, data types, and value formats of the conversion command are changed to conform to the protocol syntax requirements of the target channel. The converted commands are then syntax-validated to ensure there are no syntax conflicts or format errors. All validated commands are then integrated into a set to generate an intermediate protocol command set.

[0115] Among them, the intermediate protocol instruction set is a set of policy instructions that, after being converted by the protocol syntax, conform to the target channel protocol syntax requirements, but have not yet been adapted to the terminal format.

[0116] (3) Combine the user terminal context data to perform adaptive format reconstruction of the intermediate protocol instruction set to generate a channel executable instruction set.

[0117] For example, context data from the user terminal is collected, including information such as terminal hardware model, operating system version, browser type, screen resolution, and network environment type. For each instruction in the intermediate protocol instruction set, the format adaptation requirements corresponding to the terminal context data are analyzed, such as interface display layout requirements, data transmission compression standards, and instruction execution triggering methods. Based on these adaptation requirements, the intermediate protocol instructions are restructured, adjusting the content organization, data display dimensions, and interactive response logic to ensure the instructions are compatible with the terminal's operating environment and usage scenarios. Terminal compatibility testing is performed on the restructured instructions to verify their recognizability and executability on the terminal. Instructions that pass the tests are integrated into a set to generate a channel-executable instruction set.

[0118] Among them, the channel executable instruction set is a set of policy instructions that can be directly parsed and executed by the target channel and the corresponding terminal after protocol conversion and terminal format reconstruction.

[0119] (4) Perform fault-tolerant redundancy coding on the channel executable instruction set to generate a channel executable instruction set with self-verification mechanism.

[0120] For example, a data integrity verification field is added to each instruction in the channel executable instruction set, specifying the verification algorithm and logic to verify whether data loss or tampering has occurred during instruction transmission. Simultaneously, redundant data fragments are embedded, and a backup mechanism is designed for the core content of the instruction to ensure complete recovery through redundant data should partial data in the instruction be corrupted. Based on the verification field and redundant data, an instruction self-verification process is constructed, enabling the terminal to automatically complete data verification and integrity repair before executing the instruction. The instructions with added verification fields, redundant data, and self-verification processes are then encapsulated as a whole, and all encapsulated instructions are integrated into a set to generate a channel executable instruction set with a self-verification mechanism.

[0121] Among them, the channel executable instruction set with self-verification mechanism is a set of strategy instructions that adds verification and redundancy mechanisms to the channel executable instruction set, and has self-verification and repair capabilities, thereby improving the stability and reliability of instruction execution.

[0122] In one embodiment, the positioning encoding matrix is ​​subjected to policy knowledge base mapping and matching processing to generate a differentiated policy instruction set, including:

[0123] (1) Perform hierarchical bit sequence decoding on the positioning coding matrix to generate a set of value level identifiers, a set of behavior pattern identifiers, and a set of demand type identifiers.

[0124] For example, for the positioning encoding matrix, the start and end positions of the bit sequences at each level in its matrix structure are analyzed to clarify the boundaries of the value level bit sequences, behavior pattern bit sequences, and demand type bit sequences. According to a preset decoding dictionary, the value level bit sequences are converted into corresponding value level identifiers, and integrated to form a value level identifier set. The behavior pattern bit sequences are converted into corresponding behavior feature identifiers, and integrated to form a behavior pattern identifier set. The demand type bit sequences are converted into corresponding demand category identifiers, and integrated to form a demand type identifier set. The integrity of the three generated identifier sets is verified to ensure no omissions or errors in identification.

[0125] Among them, the value level identifier set consists of symbols representing customer value levels, the behavior pattern identifier set consists of symbols representing customer behavior characteristics, and the demand type identifier set consists of symbols representing customer demand categories.

[0126] (2) Based on the preset triple association rule base, the value level identifier set, behavior pattern identifier set and demand type identifier set are jointly retrieved and processed to generate an initial strategy matching set.

[0127] For example, based on a pre-defined triplet association rule base, which contains policy entries indexed by combinations of value hierarchy identifiers, behavioral pattern identifiers, and demand type identifiers, each index corresponds to at least one applicable policy. Identifiers corresponding to the customer group are selected from the value hierarchy identifier set, behavioral pattern identifier set, and demand type identifier set, and combined to form a unique triplet search key. The rule base is traversed according to the triplet search key, matching all associated policy entries and recording information such as the policy objective, execution method, and resource configuration contained in each entry. A duplicate check is performed on the matched policy entries, completely duplicate entries are removed, unique policy content is retained, and these are integrated to form a set, generating an initial policy matching set.

[0128] The preset triple association rule base is a set of rules that stores the association relationship between identifier combinations and policy entries, and the initial policy matching set is a set of policy entries associated with triple retrieval keys.

[0129] (3) The initial policy matching set is dynamically optimized by integrating real-time scene parameters to generate an enhanced policy instruction set.

[0130] For example, real-time scenario parameters are collected, including the customer's current marketing stage, the load status of interactive channels, the customer's historical response time preferences, and the real-time promotional campaign theme. Pre-defined applicable scenario conditions for each strategy item in the initial strategy matching set are extracted, including applicable time periods, channel restrictions, and campaign compatibility.

[0131] The system compares real-time scenario parameters with preset applicable scenario conditions across multiple dimensions to calculate a matching score. Based on the score, the execution priority of strategy items is adjusted. Strategy items with low matching scores are supplemented with scenario-adaptive content, such as adjusting the wording style or optimizing the display format. The adjusted strategy items undergo scenario adaptability verification to ensure they meet real-time scenario requirements. Items that pass verification are then integrated into a set to generate an enhanced strategy instruction set.

[0132] Among them, the enhanced policy instruction set is a set of policy instructions that has been optimized for real-time scene parameters and has scene adaptability.

[0133] (4) Perform policy conflict detection and compliance filtering on the enhanced policy instruction set to generate a differentiated policy instruction set.

[0134] For example, for the enhanced strategy instruction set, key attributes such as the target customer scope, execution time window, resource allocation, and core marketing content of each strategy instruction are extracted. A strategy conflict detection algorithm is used to compare the key attributes of each instruction and identify execution conflicts caused by attribute overlap, such as situations where the same customer scope is allocated excessive resources in the same time window. Conflicting instructions are sorted according to business importance, retaining high-priority instructions or merging conflicting content.

[0135] Based on industry compliance standards, data privacy regulations, and corporate business practices, we construct compliance filtering rules to check the legality of each policy instruction's content and its suitability for customer permissions. Instructions that violate compliance rules are filtered out, and instructions suspected of having compliance risks are revised and improved. Instructions that have undergone conflict resolution and compliance filtering are integrated to ensure that each instruction is targeted at a specific customer group and has unique content, generating a differentiated set of policy instructions.

[0136] Among them, the differentiated strategy instruction set is a set of personalized strategy instructions that are tailored to different customer positioning and have undergone conflict handling and compliance filtering.

[0137] In one embodiment, the distribution effectiveness quantification parameters are obtained using the following formula:

[0138]

[0139] in, This represents the quantification parameter of distribution effectiveness. This represents the efficiency quantification function. Indicates the number of customer touchpoint systems. Represents the message bus state matrix. This represents the load balancing scheduling function. This represents protocol conversion and format encapsulation functions. Indicates the set of distribution paths to be adapted. This represents the instruction set for differentiated strategies.

[0140] For example, through protocol conversion and format encapsulation functions Differentiated strategy instruction set According to the adapted distribution path set The channel protocol requires format conversion and encapsulation. This is handled by the load balancing scheduling function. Combined with message bus state matrix For the converted instructions Load balancing scheduling is performed among individual customer touchpoint systems. This is achieved through efficiency quantification functions. The scheduling results of all customer touchpoint systems are summarized and quantified to obtain distribution efficiency quantification parameters that reflect the omnichannel distribution effectiveness of strategy instructions. .

[0141] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0142] In one embodiment, such as Figure 3 As shown, this application also provides a big data-based precision marketing customer positioning device 300, which includes:

[0143] The profile integration module 301 is used to integrate heterogeneous data from the customer transaction database, behavior log database and attribute knowledge base to generate a panoramic customer profile dataset.

[0144] Clustering coding module 302 is used to perform customer segmentation processing based on panoramic customer profile dataset and generate location coding matrix through a preset multi-level clustering engine.

[0145] The strategy mapping module 303 is used to perform strategy knowledge base mapping and matching processing on the positioning encoding matrix to generate a differentiated strategy instruction set;

[0146] The channel distribution module 304 is used to distribute the differentiated strategy instruction set across all channels and output it to the customer touchpoint system for execution.

[0147] Specifically, the profiling integration module 301 cleans the transaction record data in the customer transaction database, the behavioral trajectory data in the behavior log database, and the customer attribute data in the attribute knowledge base, removing invalid records and duplicate content, and standardizing data field names and formats. It establishes cross-database relationships based on the customer's unique identifier, eliminating logical conflicts between heterogeneous data. Through data fusion technology, it integrates the feature dimensions of the three types of data to form a complete dataset covering transaction features, behavioral features, and attribute features, generating a panoramic customer profile dataset.

[0148] Among them, the panoramic customer profile dataset is a dataset that integrates multi-dimensional information on customer transactions, behaviors and attributes, comprehensively reflecting customer characteristics.

[0149] The clustering coding module 302 extracts customer value features, behavioral features, and demand features from the panoramic customer profile dataset and inputs them into a preset multi-level clustering engine. The first-level clustering mechanism clusters the customer value features, dividing them into different value-level groups. Within each value-level group, the second-level clustering mechanism performs deep clustering of customer behavioral features, generating behavioral feature subgroups. Based on these behavioral feature subgroups, customer demand tendencies are mined and demand labels are generated. The value labels of the value-level groups, the behavioral labels of the behavioral feature subgroups, and the demand labels are combined and encoded according to preset rules to form a matrix structure, generating a positioning coding matrix.

[0150] The positioning coding matrix is ​​a matrix that integrates three types of tag codes: customer value, behavior, and needs, and is used to identify customer positioning.

[0151] The strategy mapping module 303 decodes the positioning encoding matrix, extracting value level identifiers, behavioral pattern identifiers, and demand type identifiers. Based on a preset triplet association rule base, it combines these three types of identifiers to form search conditions, retrieves matching strategy entries, and generates an initial strategy matching set. The initial strategy matching set is optimized by integrating real-time scenario parameters, adjusting strategy priorities and supplementing scenario-adaptive content. The optimized strategies undergo conflict detection and compliance filtering, eliminating conflicting and non-compliant strategies, and integrating them to form a personalized instruction set for different customer positioning, generating a differentiated strategy instruction set.

[0152] Among them, the differentiated strategy instruction set is a set of personalized strategy instructions that have been optimized for scenarios, conflict handling, and compliance filtering.

[0153] The channel distribution module 304 performs compatibility analysis on the differentiated strategy instruction set and the technical characteristics and terminal requirements of each marketing channel to determine the appropriate distribution path and generate an appropriate distribution path set. Based on the appropriate distribution path set, the strategy instructions are protocol-converted and format-encapsulated to meet the execution requirements of the channels and terminals, generating a channel-executable instruction set. The channel-executable instruction set is scheduled through the load balancing mechanism of the asynchronous message bus to balance the load pressure of each customer touchpoint system. The scheduled instructions are output to the customer touchpoint system, which executes the operation according to the instruction requirements and collects execution feedback data.

[0154] Among them, the customer touchpoint system is the terminal system that receives and executes channel-executable instructions, and is responsible for the implementation of policy instructions.

[0155] Clustering encoding module 302 is also used for:

[0156] The following formula is used to perform joint clustering of contribution and activity levels on the panoramic customer profile dataset to generate value-level grouping sets:

[0157]

[0158] in, Represents a value hierarchy grouping set. Indicates the first Each customer's contribution feature vector Indicates the first The contribution cluster centers of the group Indicates the first Customer activity feature vector Indicates the first The activity cluster centers of the group This represents the contribution weighting coefficient. This represents the minimization optimization operator. express A set of groups, Indicates the first A collection of customers at different value levels;

[0159] Based on the value hierarchy grouping set, deep learning processing of interaction trajectories is performed within each value hierarchy group to generate behavioral feature sub-clusters;

[0160] Perform response pattern correlation analysis on behavioral feature subclusters to generate a set of demand dimension labels;

[0161] The value tags of the value hierarchy group set, the behavior tags of the behavior feature subgroup, and the demand tags of the demand dimension tag set are combined and encoded to generate a positioning encoding matrix.

[0162] Clustering encoding module 302 is also used for:

[0163] Multi-dimensional feature decoupling processing is performed on user interaction trajectories within the value hierarchy group set to generate a spatiotemporal sequence feature set.

[0164] Deep pattern extraction is performed on the spatiotemporal sequence feature set using an attention-enhanced recurrent neural network to generate a set of behavioral prototype vectors.

[0165] Dynamic density clustering is performed based on semantic similarity in the behavioral prototype vector set to generate behavioral feature sub-clusters.

[0166] Interactive intent reasoning is performed on the behavioral feature subgroups to generate behavioral feature subgroups with semantic labels.

[0167] Channel distribution module 304 is also used for:

[0168] Perform channel compatibility analysis on the differentiated strategy instruction set to generate an adapted distribution path set;

[0169] Based on the adapted distribution path set, the policy instructions are processed by protocol conversion and format encapsulation to generate a channel executable instruction set;

[0170] The executable instruction set of the channel is load-balanced and scheduled through an asynchronous message bus, and then output to the customer touchpoint system for execution.

[0171] Collect execution feedback data from customer touchpoint systems to generate quantitative parameters for distribution effectiveness.

[0172] Channel distribution module 304 is also used for:

[0173] Perform channel protocol feature parsing on the adapted distribution path set to generate a protocol feature descriptor set;

[0174] Based on the protocol feature descriptor set, an intermediate protocol instruction set is generated by performing instruction syntax conversion processing through a preset dynamic protocol conversion rule base.

[0175] The intermediate protocol instruction set is adaptively restructured by combining user terminal context data to generate a channel executable instruction set.

[0176] The channel executable instruction set is subjected to fault-tolerant redundancy coding processing to generate a channel executable instruction set with a self-verification mechanism.

[0177] The policy mapping module 303 is also used for:

[0178] The positioning encoding matrix is ​​subjected to hierarchical bit sequence decoding to generate a set of value level identifiers, a set of behavior pattern identifiers, and a set of demand type identifiers.

[0179] Based on the preset triple association rule base, the value level identifier set, behavior pattern identifier set and demand type identifier set are jointly retrieved and processed to generate an initial strategy matching set.

[0180] The initial policy matching set is dynamically optimized by integrating real-time scene parameters to generate an enhanced policy instruction set;

[0181] Perform policy conflict detection and compliance filtering on the enhanced policy instruction set to generate a differentiated policy instruction set.

[0182] Channel distribution module 304 is also used for:

[0183]

[0184] in, This represents the quantification parameter of distribution effectiveness. This represents the efficiency quantification function. Indicates the number of customer touchpoint systems. Represents the message bus state matrix. This represents the load balancing scheduling function. This represents protocol conversion and format encapsulation functions. Indicates the set of distribution paths to be adapted. This represents the instruction set for differentiated strategies.

[0185] In one embodiment, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0186] In one embodiment, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above-described method embodiments.

[0187] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0188] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A big data-based precision marketing customer positioning method, characterized in that: The method includes: Heterogeneous data integration and processing are performed on the customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset. Based on the panoramic customer profile dataset, a multi-level clustering engine is used to perform customer segmentation processing to generate a location coding matrix. The positioning encoding matrix is ​​subjected to policy knowledge base mapping and matching processing to generate a differentiated policy instruction set; The differentiated strategy instruction set is distributed across all channels and output to the customer touchpoint system for execution.

2. The method for precise marketing customer positioning based on big data according to claim 1, characterized in that, The process of generating a location coding matrix based on the panoramic customer profile dataset, using a preset multi-level clustering engine to perform customer segmentation, includes: The following formula is used to perform contribution-activity joint clustering on the panoramic customer profile dataset to generate a value-level grouping set: in, Represents a value hierarchy grouping set. Indicates the first Each customer's contribution feature vector Indicates the first The contribution cluster centers of the group Indicates the first Customer activity feature vector Indicates the first The activity cluster centers of the group This represents the contribution weighting coefficient. This represents the minimization optimization operator. express A set of groups, Indicates the first A collection of customers at different value levels; Based on the value hierarchy grouping set, deep learning processing of interaction trajectories is performed within each value hierarchy group to generate behavioral feature sub-clusters. The behavioral feature sub-clusters are subjected to response pattern association analysis to generate a demand dimension label set; The value tags of the value hierarchy group set, the behavior tags of the behavior feature subgroup, and the demand tags of the demand dimension tag set are combined and encoded to generate the positioning encoding matrix.

3. The method for precise marketing customer positioning based on big data according to claim 2, characterized in that, The step of performing deep learning processing on interaction trajectories within each value-level group based on the value-level grouping set to generate behavioral feature sub-clusters includes: Multi-dimensional feature decoupling processing is performed on the user interaction trajectories within the value hierarchy group set to generate a spatiotemporal sequence feature set. Deep pattern extraction is performed on the spatiotemporal sequence feature set using an attention-enhanced recurrent neural network to generate a set of behavioral prototype vectors. Dynamic density clustering is performed based on the semantic similarity of the behavioral prototype vector set to generate the behavioral feature sub-clusters; The behavioral feature subgroups are subjected to interactive intent reasoning processing to generate the behavioral feature subgroups with semantic labels.

4. The method for precise marketing customer positioning based on big data according to claim 1, characterized in that, The step of distributing the differentiated strategy instruction set across all channels and outputting it to the customer touchpoint system for execution includes: The differentiated strategy instruction set is subjected to channel compatibility analysis and processing to generate an adapted distribution path set; Based on the adapted distribution path set, the policy instructions are processed by protocol conversion and format encapsulation to generate a channel executable instruction set. The executable instruction set of the channel is load-balanced and scheduled through an asynchronous message bus, and then output to the customer touchpoint system for execution. Collect execution feedback data from the customer touchpoint system to generate distribution efficiency quantification parameters.

5. The method for precise marketing customer positioning based on big data according to claim 4, characterized in that, The step of performing protocol conversion and format encapsulation processing on the policy instructions based on the adapted distribution path set to generate a channel-executable instruction set includes: The adapted distribution path set is subjected to channel protocol feature parsing processing to generate a protocol feature descriptor set; Based on the protocol feature descriptor set, an intermediate protocol instruction set is generated by performing instruction syntax conversion processing through a preset dynamic protocol conversion rule base. The intermediate protocol instruction set is adaptively restructured by combining user terminal context data to generate the channel executable instruction set. The channel executable instruction set is subjected to fault-tolerant redundancy encoding processing to generate the channel executable instruction set with a self-verification mechanism.

6. The method for precise marketing customer positioning based on big data according to claim 1, characterized in that, The step of performing policy knowledge base mapping and matching processing on the positioning encoding matrix to generate a differentiated policy instruction set includes: The location encoding matrix is ​​subjected to hierarchical bit sequence decoding to generate a value level identifier set, a behavior pattern identifier set, and a demand type identifier set. Based on a preset triplet association rule base, the value hierarchy identifier set, the behavior pattern identifier set, and the demand type identifier set are jointly retrieved to generate an initial strategy matching set. The initial policy matching set is dynamically optimized by integrating real-time scene parameters to generate an enhanced policy instruction set; The enhanced policy instruction set is subjected to policy conflict detection and compliance filtering to generate the differentiated policy instruction set.

7. The method for precise marketing customer positioning based on big data according to claim 4, characterized in that, The distribution efficiency quantification parameters are obtained using the following formula: in, This represents the quantification parameter of distribution effectiveness. This represents the efficiency quantification function. Indicates the number of customer touchpoint systems. Represents the message bus state matrix. This represents the load balancing scheduling function. This represents protocol conversion and format encapsulation functions. Indicates the set of distribution paths to be adapted. This represents the instruction set for differentiated strategies.

8. A big data-based precision marketing customer positioning device, characterized in that: The device includes: The profile integration module is used to integrate heterogeneous data from the customer transaction database, behavior log database, and attribute knowledge base to generate a panoramic customer profile dataset. The clustering coding module is used to perform customer segmentation processing based on the panoramic customer profile dataset and generate a location coding matrix by using a preset multi-level clustering engine. The strategy mapping module is used to perform strategy knowledge base mapping and matching processing on the positioning encoding matrix to generate a differentiated strategy instruction set; The channel distribution module is used to distribute the differentiated strategy instruction set across all channels and output it to the customer touchpoint system for execution.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the big data-based precision marketing customer positioning method as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the big data-based precision marketing customer positioning method as described in any one of claims 1 to 7.

Images