Big data based short message transmission optimization system

Abstract

This invention discloses a big data-based SMS sending optimization system, belonging to the field of big data marketing technology. The system includes: a compliance pre-verification module, a multi-version content generation module, a spatiotemporal test matrix construction module, a dynamic traffic allocation module, and a strategy decision feedback module. In terms of user grouping and traffic allocation, this invention uses user profile feature screening and Pearson correlation coefficient to accurately select features highly correlated with SMS response rate, dividing the target user group into K homogeneous groups and rigorously verifying the internal homogeneity and inter-group heterogeneity of the groups. On this basis, each group of users is mapped to a test unit matrix proportionally, and the consistency between the sample feature distribution and the target group is verified by calculating feature statistics. This ensures that the samples of the test units represent the target group to the greatest extent, making the data relied upon by the subsequent strategy decision feedback module more representative and scientific.

Description

Technical Field

[0001] This invention relates to the field of big data marketing technology, specifically to a big data-based SMS sending optimization system. Background Technology

[0002] In today's digital age, communication methods between businesses and users are becoming increasingly diverse. SMS, as a basic and important means of information delivery, continues to play a crucial role in many fields. Whether it's e-commerce platforms pushing promotional activities to users, financial institutions informing customers of account changes, or educational institutions sending course notifications, SMS has become an effective channel for information dissemination due to its immediacy and convenience. With the rise of big data technology, how to optimize SMS sending by leveraging massive amounts of data to improve user response rates and marketing effectiveness has become a key issue that businesses urgently need to address.

[0003] However, current technologies for optimizing SMS delivery based on big data have significant shortcomings. Many systems fail to deeply consider the intrinsic relationship between user profile characteristics and SMS response rates when grouping users, simply relying on a few superficial features for grouping. This results in insufficient homogeneity and heterogeneity of the groups, making it impossible to accurately distinguish the differences in SMS responses among different user groups. In the traffic allocation stage, there is a lack of scientific mapping and verification mechanisms. The feature distribution of test unit samples is often disconnected from the target group, making subsequent strategies derived from test samples lack universality and effectiveness. Furthermore, existing strategy decision feedback modules often neglect the analysis of the complex interaction effects between content, time, and user grouping, making it difficult to formulate accurate and effective SMS delivery strategies for different user groups.

[0004] In conclusion, developing a big data-based SMS sending optimization system that can accurately segment user groups, scientifically allocate traffic, and deeply analyze the effects of strategies to achieve precise feedback and optimization is of significant practical importance. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and provide a big data-based SMS sending optimization system. This system, in user grouping and traffic allocation, uses user profile feature filtering and Pearson correlation coefficient to accurately select features highly correlated with SMS response rates. It divides the target user group into K homogeneous groups and rigorously verifies the internal homogeneity and inter-group heterogeneity of these groups. Based on this, each group of users is mapped to a test unit matrix at a 5% ratio, and the consistency between the sample feature distribution and the target group is verified by calculating feature statistics. This ensures that the test unit samples represent the target group to the greatest extent possible, providing a reliable data foundation for subsequent strategy formulation. This makes the data relied upon by the subsequent strategy decision feedback module more representative and scientific, ultimately making the SMS sending strategies derived from this data more suitable for the needs of different user groups and improving the overall SMS marketing effect.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a big data-based SMS sending optimization system, the components of which include: a compliance pre-verification module, a multi-version content generation module, a spatiotemporal test matrix construction module, a dynamic traffic allocation module, and a strategy decision feedback module;

[0007] The compliance pre-verification module is used to connect to the compliance database, perform compliance pre-verification on all versions of SMS content, and intercept SMS content versions containing prohibited words and false advertising through real-time semantic analysis, and output a compliance content pool.

[0008] The multi-version content generation module generates N SMS content versions based on user historical behavior data, and inputs them into the compliance pre-verification module for filtering to generate the final test content set. ;

[0009] The spatiotemporal test matrix construction module combines the test content set C with the sending time set. Perform Cartesian product combinations to construct the test unit matrix. Each unit Content In time The sending strategy;

[0010] The dynamic traffic allocation module divides the user group into K groups based on user profile characteristics, and maps each group to different test units of the CT matrix according to a preset ratio to ensure that the test samples of each unit are evenly distributed.

[0011] The strategy decision feedback module performs quantitative analysis of various... The conversion rate data of the unit is used to decouple the independent effects of content factors and time factors, output the optimal content-time combination strategy, and feed it back to the multi-version content generation module for iterative optimization.

[0012] Furthermore, the compliance pre-verification module includes:

[0013] Compliance Database Interface Unit: Used for real-time connection to a compliance database that includes a library of prohibited words in advertising law, a library of industry standard clauses, and a library of false advertising characteristics, and supports automatic updates and synchronization of the database;

[0014] Semantic analysis engine: Employs a Bi-LSTM (Bi-Long Short-Term Memory) network to segment the input SMS content, extracting semantic feature vectors. These vectors are then compared with feature vectors in a compliance database to calculate the violation confidence score for risky phrases. Identify potential banned words and false advertising statements in the content;

[0015] The violation confidence score ,in, The number of key phrases extracted from the text message content. For the first The Bi-LSTM output feature vector of each key phrase For the first in the compliance database The standard feature vector of each violation feature, The cosine similarity function is used. Words that violate regulations Weighting in historical violation cases The weighting coefficient is adjustable. And default value , This represents the maximum similarity to all violation features.

[0016] Furthermore, the compliance pre-verification module also includes a tiered interception unit, which performs tiered operations based on the violation confidence score:

[0017] when If a high-risk violation is detected, the system will immediately block the violation, generate an alert, and trigger a manual review process.

[0018] when If a general violation is found, the text message content is rewritten, and the confidence score is recalculated after rewriting.

[0019] when At that time, content is allowed to enter the compliant content pool;

[0020] The hierarchical interception unit records in detail the content version, interception result, violation confidence score, processing method and timestamp for each round of pre-verification, forming a traceable compliance verification log, and feeding back frequently occurring violation characteristics to the compliance database.

[0021] Furthermore, the multi-version content generation module includes:

[0022] User Feature Extraction Unit: Used to acquire multi-dimensional user data from the big data platform and construct user feature vectors. in, Based on historical behavioral characteristics, the user feature vectors are weighted and fused to obtain a comprehensive user feature score. in, The feature weights are those that satisfy the following conditions: ;

[0023] Content Template Library: Stores scenario-based content templates categorized by industry scenarios. Each template contains fixed text segments and variable placeholders, and is associated with speech style tags.

[0024] Generative AI Engine: Employs a lightweight generative model, taking user feature vectors as input. Combine content templates to generate personalized content candidates;

[0025] Version diversity verification: Based on the basic content output by the generative AI engine, an initial version is generated by synonym replacement, sentence transformation, and addition / removal of redundant information. The semantic similarity between the two initial versions is then calculated. The output value range is [0, 1]. N target versions are selected from the initial version to ensure that the average similarity between versions satisfies [0, 1]. ,in, Indicates from The number of combinations of selecting two elements from n elements without order. For diversity threshold, Representing the Each version of the SMS content Representing the For each SMS content version, the iteration count is automatically increased if the filtered results do not meet the threshold, outputting N personalized SMS content versions that have passed diversity validation. These are then input into the compliance pre-validation module for filtering, and after removing non-compliant versions, the final test content set is generated. ,in and To ensure the validity of the test, This represents the number of items retained after filtering by the compliance pre-verification module. Each version of the SMS message content.

[0026] Furthermore, the specific process by which the generative AI engine generates personalized content candidates is as follows:

[0027] Template matching: Score users based on their overall characteristics The templates are matched with the applicable feature ranges of the template library, and the three candidate templates with the highest matching degree are selected.

[0028] Variable population: User features are associated with template variables based on entity linking technology. This entity linking technology performs type parsing on template variable placeholders, establishes a variable type library, and assigns a unique ID to each variable type from the user feature vector. Extract entities that match the type, construct an entity pool, and calculate the semantic association between entities and variables using the following formula: ,in, Candidate entities in the entity pool For template variables, For BERT embedding vectors of entities / variables, To populate the variable with the entity that appears most frequently in the user's historical behavior, select the entity with the highest Rel value.

[0029] Style Adjustment: Based on the wording style tags in the user tags, the template text is rewritten to adapt to the user and obtain personalized content candidates.

[0030] Furthermore, the spatiotemporal test matrix construction module includes:

[0031] Time Set Generation Unit: Used to extract user activity time features from user historical behavior data and generate a set of sending times. ;

[0032] Matrix building unit: the final set of test content With the set of sending times Combine them to construct a test unit matrix. Each unit Content In time The sending strategy;

[0033] The output of the spatiotemporal test matrix construction module is a dynamically adjusted test unit matrix. Each unit A corresponding content-time combination strategy is input into the dynamic traffic allocation module for sample allocation.

[0034] Furthermore, the dynamic traffic allocation module includes: a user grouping unit, a traffic mapping unit, and a traffic balance verification unit, used to allocate user groups to different units of the test unit matrix, wherein:

[0035] User grouping unit: Based on user profile features, from user feature vectors The target user group was divided into K homogeneous groups based on features highly correlated with SMS response rate. The correlation coefficient of SMS response rate was Pearson correlation coefficient. Filter features, calculate the internal homogeneity and inter-group heterogeneity of each group, and verify the effectiveness of the grouping. The internal homogeneity should satisfy the variance of the SMS response rate of users within the group ≤ 0.1, and the inter-group heterogeneity should satisfy the difference in the mean of the SMS response rate between groups ≥ 0.2.

[0036] Traffic mapping unit: Maps users from each group to the test unit matrix at a ratio of 5%. Different units;

[0037] Balance verification unit: Verifies the consistency between the sample feature distribution of each test unit and the target population, calculates the feature statistics of each test unit sample, and evaluates the degree of difference between the sample feature distribution and the target population. This degree of difference is obtained by calculating the standard deviation of the test unit sample and the target population. Then adjust the traffic allocation ratio and reallocate the samples;

[0038] The output of the dynamic traffic allocation module is a uniformly distributed test sample, with each test unit... A set of homogeneous user samples is input into the strategy decision feedback module for effect analysis.

[0039] Furthermore, the strategy decision feedback module includes:

[0040] Group Response Analysis Unit: For the K homogeneous user groups divided by the dynamic traffic allocation module, the response data of each test unit in different groups is statistically analyzed. The response data includes the open rate. Conversion rate The opening rate This indicates that the content in the k-th user group is... In time The percentage of users who opened the text message after sending it, the conversion rate This represents the percentage of users in the k-th user group who completed the target action after opening the SMS message. A three-dimensional response matrix of group, content, and time is constructed, where the value of each cell in the matrix is... ;

[0041] Causal decoupling unit: Separates the independent effects of content and time on user response, and calculates the combined effect score. Regarding the content independence effect Considering only the content The impact on the k-th group, excluding the interference of time factors, is calculated by taking time into account. Fixed as a constant Statistical content The conversion rate during this period, i.e. Time independence effect Considering only time The impact on the k-th group, excluding content-related interference, is calculated by taking the content... Fixed to the content with the highest historical response rate for this user group. Statistics on this fixed content over time The conversion rate, i.e. The combined effect score ;

[0042] Grouping Optimal Strategy for Unit Selection: For the k-th group, select all test units... Sort the test units from highest to lowest and select the top 10% as the optimal strategy. Feedback is then fed back to the multi-version content generation module for iterative optimization, guiding it to generate content containing this feature.

[0043] Compared with existing technologies, this big data-based SMS sending optimization system has the following advantages:

[0044] I. In user grouping and traffic allocation, this invention uses user profile feature screening and Pearson correlation coefficient to accurately select features highly correlated with SMS response rate, dividing the target user group into K homogeneous groups. It rigorously verifies the internal homogeneity and inter-group heterogeneity of these groups. Based on this, each group of users is mapped to a test unit matrix at a 5% ratio. The consistency between the sample feature distribution and the target group is verified by calculating feature statistics. This ensures that the test unit samples represent the target group to the greatest extent possible, providing a reliable data foundation for subsequent strategy formulation. This makes the data relied upon by the subsequent strategy decision feedback module more representative and scientific, ultimately making the SMS sending strategy derived from this data more suitable for the needs of different user groups and improving the overall SMS marketing effect.

[0045] Second, this invention constructs a three-dimensional response matrix of group-content-time through a group response analysis unit, comprehensively recording the response of different groups to different content and sending times. The causal decoupling unit accurately separates the independent influence of content and time factors on user response, and then calculates the comprehensive effect score. The strategy screening unit selects and verifies the optimal content-time combination for each group based on the comprehensive effect score, transforms these optimal strategies into rules, and feeds them back to the multi-version content generation module for continuous iteration and optimization. This can continuously improve the accuracy and effectiveness of SMS sending strategies, helping enterprises to better utilize SMS for precise marketing and improve user response rate and conversion rate.

[0046] Other advantages, objectives and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination or study, or may be learned from the practice of the invention. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 Flowchart of the operation of the SMS sending optimization system based on big data;

[0049] Figure 2 A diagram showing the module composition of a big data-based SMS sending optimization system;

[0050] Figure 3 This is a schematic diagram illustrating the working principle of the generative AI engine in Example 1. Detailed Implementation

[0051] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1

[0053] This embodiment aims to explain in detail the working principle of a big data-based SMS sending optimization system, such as... Figure 2 As shown, this system achieves precise optimization of SMS sending through the collaborative work of a compliance pre-verification module, a multi-version content generation module, a spatiotemporal test matrix construction module, a dynamic traffic allocation module, and a strategy decision feedback module. Starting with compliance verification of SMS content, the system generates multiple versions of personalized content, constructs a test matrix based on the time dimension, conducts tests through scientific user grouping and traffic allocation, and finally determines the optimal sending strategy based on feedback data and continuously iterates, effectively improving the response rate and conversion rate of SMS marketing while ensuring the compliance of SMS content.

[0054] In its implementation, the compliance pre-verification module interfaces with a compliance database to pre-verify the compliance of all versions of SMS content, blocking content containing prohibited words and false advertising, and outputting a compliance content pool to provide a legal and compliant content foundation for subsequent SMS sending. This module mainly consists of a compliance database interface unit, a semantic analysis engine, and a tiered interception unit. The compliance database interface unit interfaces with the compliance database in real time to ensure the system can obtain the latest compliance standards promptly. The compliance database contains key content such as a prohibited advertising word library, an industry standard clause library, and a false advertising feature library. To ensure the timeliness of the database information, the interface unit supports automatic database updates. The system employs a new synchronization mechanism. When industry standards change, the database updates rapidly, and the interface unit detects and synchronizes these changes to the system in real time, ensuring that content verification always adheres to the latest compliance requirements. The semantic analysis engine uses a Bidirectional Long Short-Term Memory (Bi-LSTM) network to process the input SMS content. First, the SMS content is segmented into meaningful words or phrases. Then, Bi-LSTM extracts semantic feature vectors from these words or phrases, accurately representing the semantic information of the text. To determine whether the SMS content violates regulations, a violation confidence score for the risk phrases needs to be calculated. The calculation formula is as follows: ,in, The number of key phrases extracted from the text message content. For the first The Bi-LSTM output feature vector of each key phrase For the first in the compliance database The standard feature vector of each violation feature, is the cosine similarity function, used to calculate the similarity between two feature vectors. The closer the value is to 1, the more semantically similar the two are. Words that violate regulations The weighting of the term in historical violation cases reflects its importance in violation determination; and The weighting coefficients are adjustable, and ,default value This is used to balance the influence of semantic similarity and the weight of violating words in the violation confidence score; This means taking the maximum similarity among all violation features. Specifically, for each key phrase, the most similar violation feature is found, and the result of this most similar feature's calculation is used as the contribution of that key phrase to the violation confidence score. The violation confidence score is calculated using this formula. It can comprehensively consider the semantic similarity between key phrases and violation features in SMS content, as well as the historical weight of violation words, thereby accurately identifying potential prohibited words and false advertising statements in the content; the hierarchical interception unit calculates the violation confidence score based on the semantic analysis engine. Perform hierarchical operations: when When a high-risk violation is detected, the system will immediately block the SMS message and generate an alert, triggering a manual review process. This is because a score reaching this threshold indicates a very high probability of a serious violation, requiring further detailed manual review to prevent the message from being sent. If a minor violation is detected, the system will automatically rewrite the SMS content and recalculate the confidence score. This is because such violations are relatively minor, and automatic rewriting may bring them to compliance standards. The recalculation is to confirm whether the rewritten content meets the requirements. When the SMS content is deemed compliant, it indicates a high level of compliance, allowing the content to enter the compliant content pool for subsequent testing and sending. Simultaneously, the tiered interception unit meticulously records the content version, interception results, violation confidence score, processing method, and timestamp for each round of pre-verification, forming a traceable compliance verification log. These logs not only provide a basis for subsequent compliance audits but also feed back frequently occurring violation characteristics to the compliance database, further enriching and improving the compliance database and enhancing the system's compliance verification capabilities.

[0055] The multi-version content generation module generates multiple diverse SMS content versions based on user historical behavior data, which are then input into the compliance pre-verification module for filtering, ultimately generating a test content set. This module consists of a user feature extraction unit, a content template library, a generative AI engine, and a version diversity verification unit. The user feature extraction unit obtains multi-dimensional user data from a big data platform, covering aspects such as basic user information, consumption habits, browsing history, and historical purchase behavior. Based on this data, a user feature vector is constructed. ,in To comprehensively evaluate user characteristics, a weighted fusion method is used to obtain a comprehensive user characteristic score, based on historical behavioral features. The formula is as follows: ,in, The feature weights are those that satisfy the following conditions: The feature weights are determined based on the degree of influence of each feature on the SMS response rate. The greater the influence of a feature on the SMS response rate, the higher its weight, to ensure that the user's comprehensive feature score can accurately reflect the user's potential response to SMS messages. The content template library stores scenario-based content templates categorized by industry. Each template contains a fixed text segment and variable placeholders, and is associated with a speech style tag. Different industry scenarios (such as e-commerce promotions, financial notifications, educational course promotions, etc.) have different templates. The fixed text segment ensures the professionalism and basic framework of the SMS content, while the variable placeholders provide space for personalized content filling. The speech style tag helps to adjust the style later based on user characteristics. The generative AI engine adopts a lightweight generative model, inputting user feature vectors. Combined with content templates, it generates personalized content candidates, such as Figure 3 As shown, the specific process is as follows: The user's comprehensive feature score is calculated... The template library is matched against applicable feature ranges. Each template has its corresponding applicable feature range. The three candidate templates with the highest matching degree are selected through matching to provide a foundation for subsequent content generation. Based on entity linking technology, user features are associated with template variables. First, the template variable placeholders are parsed to establish a variable type library and assign a unique ID to each variable type. Then, user feature vectors are used to generate the template variables. Extract entities with matching types and build an entity pool. To select the most suitable entity to populate the variables, it is necessary to calculate the semantic association between the entity and the variable. The formula is: ,in, Candidate entities in the entity pool For template variables, These are BERT embedding vectors for entities / variables. The BERT model can transform entities and variables into vectors with semantic information. To populate variables based on the frequency of entity occurrences in user history, entities with the highest Rel values ​​are selected to better suit users' personalized needs. The template text is rewritten to adapt to user language style tags. Initial versions are generated from the basic content output by the generative AI engine using synonym replacement, sentence structure transformation, and addition / removal of redundant information. To ensure sufficient diversity between versions, the semantic similarity between two initial versions needs to be calculated. Its output value ranges from [0, 1]. The closer the value is to 0, the greater the semantic difference between the two versions. The value is selected from the initial version. There are several target versions, such that the average similarity between versions satisfies: ,in, Indicates from The number of combinations of selecting two elements from n elements without order. For diversity threshold, Representing the Each version of the SMS content Representing the Each SMS message content version, this formula guarantees the selection. The versions are sufficiently different. When the filtering results do not meet the threshold, the system will automatically increase the number of iterations, regenerate and filter versions, until the output is complete. These personalized SMS content versions, after undergoing diversity verification, are input into the compliance pre-verification module for filtering. After removing non-compliant versions, the final test content set is generated. ,in and To ensure the validity of the test, The first one retained after being filtered by the compliance pre-verification module Each version of the SMS message content.

[0056] The spatiotemporal test matrix construction module combines the test content set and the sending time set using a Cartesian product to construct a test unit matrix, providing a structural framework for subsequent traffic allocation and strategy testing. This module includes a time set generation unit and a matrix construction unit. The time set generation unit extracts user activity time features from historical user behavior data. This historical behavior data (such as the time a user opens an SMS message, the time a user clicks a link, etc.) reflects the user's activity in different time periods. Through analysis and mining of this data, a sending time set is generated. These time points are periods when users are relatively active; sending text messages during these times can garner higher attention. The matrix construction unit will ultimately set up the test content. With the set of sending times The Cartesian product combination is performed, which means that each piece of content is paired with each point in time to construct a test unit matrix. Each unit Content In time The transmission strategy, the output of the matrix construction unit is a dynamically adjusted test unit matrix. Each unit Each content-time combination strategy corresponds to a specific strategy, which will be input into the dynamic traffic allocation module for sample allocation, providing specific strategy solutions for subsequent testing.

[0057] The dynamic traffic allocation module divides users into appropriate groups based on user profile characteristics and maps each group to different test units in the CT matrix according to a preset ratio, ensuring a balanced distribution of test samples in each unit. This module includes a user grouping unit, a traffic mapping unit, and a balance verification unit. The user grouping unit, based on user profile characteristics, selects from the user feature vector... Features highly correlated with SMS response rate were selected, as SMS response rate is an important indicator for measuring SMS effectiveness. Selecting features highly correlated with this indicator can improve the accuracy of grouping. Here, the Pearson correlation coefficient was used. To filter features, the Pearson correlation coefficient reflects the degree of linear correlation between two variables; a value greater than 0.5 indicates a strong positive correlation between the feature and the SMS response rate. Based on the filtered features, the target user group is divided into... To validate the effectiveness of a homogeneous group, it is necessary to calculate the internal homogeneity and inter-group heterogeneity of each group. Internal homogeneity requires the variance of the SMS response rate of users within the group. This means that users within the same group respond to SMS messages in a similar manner, while heterogeneity between groups requires a difference in the mean SMS response rate between groups. This ensures that there are significant differences in SMS responses among users in different groups. Such grouping provides a basis for subsequent precise strategy formulation. The traffic mapping unit maps users from each group to the test unit matrix at a ratio of 5%. The selection of a 5% proportion for different units comprehensively considers both sample representativeness and testing costs. This ensures the sample size is sufficient to reflect the overall situation without excessively increasing costs and resource consumption due to an overly large sample. The balance verification unit verifies the consistency between the sample characteristic distribution of each test unit and the target group. It assesses the difference between the sample characteristic distribution and the target group by calculating the characteristic statistics (such as mean, variance, etc.) of each test unit sample. The difference is obtained by calculating the standard deviation between the test unit sample and the target group. When the difference... If the difference is less than 0.05, it indicates a significant deviation between the sample characteristic distribution and the target group, requiring adjustment of the traffic allocation ratio and reallocation of the samples. If the difference is less than 0.05, the sample distribution is considered balanced and effective, and the output of the dynamic traffic allocation module is a balanced distribution of test samples, with each test unit... A set of homogeneous user samples will be input into the strategy decision feedback module for effect analysis, providing data support for strategy optimization.

[0058] The strategy decision feedback module quantitatively analyzes the conversion rate data of each test unit to determine the optimal content-time combination strategy and feeds it back to relevant modules for iterative optimization. This module includes a group response analysis unit, a causal decoupling unit, and a strategy filtering unit. The group response analysis unit is specifically designed for the dynamic traffic allocation module. For each homogeneous user group, the response data of each test unit in different groups is statistically analyzed, including the open rate. and conversion rate Open rate Indicates the first In each user group, the content In time Percentage of users who opened the message after sending it, conversion rate Indicates the first Within each user group, the percentage of users who completed the target action after opening the SMS message was analyzed. Based on this data, a three-dimensional response matrix of group, content, and time was constructed, where the value of each cell in the matrix is... This matrix comprehensively records the responses of different groups to different content and time factors, providing a detailed data foundation for subsequent analysis. The causal decoupling unit separates the independent effects of content factors and time factors, and calculates the comprehensive effect score. Regarding the content independence effect Considering only the content For the first The impact of grouping should be considered, excluding the interference of time factors. Time should be taken into account during calculation. Fixed as a constant (This constant is based on the first) The optimal time for determining the historical active time of grouped users, statistical content. The conversion rate during this period, i.e. Regarding the time-independent effect Considering only time For the first The impact of grouping, excluding content-related interference, and taking content into account during calculation. Fixed to the content with the highest historical response rate for this user group. Statistics on this fixed content over time The conversion rate, i.e. Overall effect score The weighting of 0.6 and 0.4 is determined based on the degree of influence of content and time factors on the final conversion rate. The strategy screening unit [is used for] the [specific parameters]. Grouping all test units into groups The top 10% of test units are selected as the optimal strategy, sorted from highest to lowest. This selection is designed to ensure the effectiveness of the strategy while retaining a number of alternative solutions for flexible adjustments based on actual circumstances. The selected optimal strategies not only include combinations of content and time but also cover corresponding user group characteristics, enabling precise matching of the needs of different user groups. These optimal strategies are fed back to the multi-version content generation module to guide subsequent content generation iterations and optimizations. Simultaneously, the strategy selection unit continuously tracks and verifies the selected optimal strategies, monitoring their actual effectiveness in real time during subsequent SMS sending. If a significant drop in conversion rate is detected, the strategy selection process is restarted to ensure that the output strategies remain optimal.

[0059] In summary, this embodiment details the complete operation process of a big data-based SMS sending optimization system. From the compliance pre-verification module ensuring the legality and compliance of SMS content through multi-dimensional compliance checks, to the multi-version content generation module generating diverse and personalized content based on user characteristics, to the spatiotemporal test matrix construction module constructing a content-time combination strategy matrix, to the dynamic traffic allocation module achieving scientific user grouping and balanced sample distribution, and finally, the strategy decision feedback module accurately analyzing and outputting the optimal strategy, a complete optimization system is formed. This system, by combining user profiles, historical behavioral data, and other multi-dimensional information, achieves intelligent operation of the entire SMS sending process, from content generation to strategy optimization.

[0060] Example 2

[0061] This embodiment provides the operation process of a big data-based SMS sending optimization system for intelligent reconciliation and verification of financial transactions, such as... Figure 1 As shown, the specific steps of this process are as follows:

[0062] (1) User profile construction and content generation

[0063] Obtain user historical behavior data from big data platforms;

[0064] Construct multi-dimensional user feature vectors (such as consumption frequency and active time periods);

[0065] Calculate the user's comprehensive feature score (weighted fusion of key features);

[0066] Match the three templates with the highest compatibility in the content template library;

[0067] Populate personalized variables (based on user characteristics and semantic relevance);

[0068] Generate basic SMS content and rewrite it to adapt the style;

[0069] N initial versions are generated through synonym replacement / sentence structure transformation;

[0070] Filter versions that meet the semantic difference standard (average similarity ≤ 0.3).

[0071] (2) Content compliance verification

[0072] Real-time connection to compliance databases (banned term database, industry standard database);

[0073] Use the Bi-LSTM model for word segmentation and semantic feature extraction;

[0074] Calculate the confidence score between the key phrases and the violation database;

[0075] Perform tiered blocking:

[0076] ≥0.8 points: High-risk and triggers manual review;

[0077] 0.3 ~ 0.8 points: Re-evaluated after automatic rewriting;

[0078] <0.3 points: Included in the compliant content pool;

[0079] Record verification logs and update the violation signature database;

[0080] (3) Construction of spatiotemporal strategy matrix

[0081] Analyze user activity periods to generate a set of sending times;

[0082] Take the Cartesian product of the compliant content pool and the time point;

[0083] Construct a two-dimensional test matrix (rows = content version, columns = sending time period);

[0084] Each matrix unit corresponds to a specific combination strategy of content and time;

[0085] (4) User grouping and traffic allocation

[0086] Select features that are strongly correlated with SMS response (Pearson coefficient > 0.5);

[0087] Divide the user group into K homogeneous groups (within-group response rate variance ≤ 0.1, between-group difference ≥ 0.2);

[0088] 5% of the users in each group will be selected.

[0089] The samples are uniformly mapped to the spatiotemporal matrix units;

[0090] Verify the consistency of sample distribution (feature difference < 0.05);

[0091] (5) Strategy Implementation and Effect Feedback

[0092] Send differentiated SMS messages based on matrix units;

[0093] Collect 3D response data (grouping × content × time);

[0094] Separation-independent effects:

[0095] Statistical analysis of content conversion rate over a fixed time period;

[0096] Conversion rate for fixed content over a specific time period;

[0097] Calculate the overall effect score (content weight 60% + time weight 40%).

[0098] Select the top 10% of the best strategies in each group;

[0099] Feedback is sent to the content generation module for iterative optimization.

[0100] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered in all respects as exemplary and non-limiting, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be included within the present invention. No reference numerals in the claims should be construed as limiting the scope of the claims.

Claims

1. A big data-based SMS sending optimization system, characterized in that: The system consists of: a compliance pre-verification module, a multi-version content generation module, a spatiotemporal test matrix construction module, a dynamic traffic allocation module, and a strategy decision feedback module; The compliance pre-verification module is used to connect to the compliance database, perform compliance pre-verification on all versions of SMS content, and intercept SMS content versions containing prohibited words and false advertising through real-time semantic analysis, and output a compliance content pool. The multi-version content generation module generates N SMS content versions based on user historical behavior data, and inputs them into the compliance pre-verification module for filtering to generate the final test content set. ; The spatiotemporal test matrix construction module combines the test content set C with the sending time set. Perform Cartesian product combinations to construct the test unit matrix. Each unit Content In time The sending strategy; The dynamic traffic allocation module divides the user group into K groups based on user profile characteristics, and maps each group to different test units of the CT matrix according to a preset ratio to ensure that the test samples of each unit are evenly distributed. The strategy decision feedback module performs quantitative analysis of various... The conversion rate data of the unit is used to decouple the independent effects of content factors and time factors, output the optimal content-time combination strategy, and feed it back to the multi-version content generation module for iterative optimization. The strategy decision feedback module includes: Group Response Analysis Unit: For the k homogeneous user groups divided by the dynamic traffic allocation module, the response data of each test unit in different groups is statistically analyzed. The response data includes the open rate. Conversion rate The opening rate This indicates that the content in the k-th user group is... In time The percentage of users who opened the text message after sending it, the conversion rate This represents the percentage of users in the k-th user group who completed the target action after opening the SMS message. A three-dimensional response matrix of group, content, and time is constructed, where the value of each cell in the matrix is... ; Causal decoupling unit: Separates the independent effects of content and time on user response, and calculates the combined effect score. Regarding the content independence effect Considering only the content The impact on the k-th group, excluding the interference of time factors, is calculated by taking time into account. Fixed as a constant Statistical content The conversion rate during this period, i.e. Time independence effect Considering only time The impact on the k-th group, excluding content-related interference, is calculated by taking the content... Fixed to the content with the highest historical response rate for this user group. Statistics on this fixed content over time Conversion rate The combined effect score ; Strategy filtering unit: For the k-th group, select all test units... Sort the test units from highest to lowest and select the top 10% as the optimal strategy. Feedback is then fed back to the multi-version content generation module for iterative optimization, guiding it to generate content containing this feature.

2. The SMS sending optimization system based on big data according to claim 1, characterized in that, The compliance pre-verification module includes: Compliance Database Interface Unit: Used for real-time connection to a compliance database that includes a library of prohibited words in advertising law, a library of industry standard clauses, and a library of false advertising characteristics, and supports automatic updates and synchronization of the database; Semantic analysis engine: Employs a Bi-LSTM (Bi-Long Short-Term Memory) network to segment the input SMS content, extracting semantic feature vectors. These vectors are then compared with feature vectors in a compliance database to calculate the violation confidence score for risky phrases. Identify potential banned words and false advertising statements in the content; The violation confidence score ,in, The number of key phrases extracted from the text message content. For the first The Bi-LSTM output feature vector of each key phrase For the first in the compliance database The standard feature vector of each violation feature, The cosine similarity function is used. Words that violate regulations Weighting in historical violation cases The weighting coefficient is adjustable. And default value , This represents the maximum similarity to all violation features.

3. The SMS sending optimization system based on big data according to claim 2, characterized in that, The compliance pre-verification module also includes a tiered interception unit, which performs tiered operations based on the violation confidence score: when If a high-risk violation is detected, the system will immediately block the violation, generate an alert, and trigger a manual review process. when If a general violation is found, the text message content is rewritten, and the confidence score is recalculated after rewriting. when At that time, content is allowed to enter the compliant content pool; The hierarchical interception unit records in detail the content version, interception result, violation confidence score, processing method and timestamp for each round of pre-verification, forming a traceable compliance verification log, and feeding back frequently occurring violation characteristics to the compliance database.

4. The SMS sending optimization system based on big data according to claim 1, characterized in that, The multi-version content generation module includes: User Feature Extraction Unit: Used to acquire multi-dimensional user data from the big data platform and construct user feature vectors. in, Based on historical behavioral characteristics, the user feature vectors are weighted and fused to obtain a comprehensive user feature score. in, The feature weights are those that satisfy the following conditions: ; Content Template Library: Stores scenario-based content templates categorized by industry scenarios. Each template contains fixed text segments and variable placeholders, and is associated with speech style tags. Generative AI Engine: Employs a lightweight generative model, taking user feature vectors as input. Combine content templates to generate personalized content candidates; Version diversity verification: Based on the basic content output by the generative AI engine, an initial version is generated by synonym replacement, sentence transformation, and addition / removal of redundant information. The semantic similarity between the two initial versions is then calculated. The output value range is [0, 1], selected from the initial version. The target versions satisfy the following average similarity between versions. ,in, Indicates from The number of combinations of selecting two elements from n elements without order. For diversity threshold, Representing the Each version of the SMS content Representing the For each SMS content version, the iteration count is automatically increased if the filtered results do not meet the threshold, outputting N personalized SMS content versions that have passed diversity validation. These are then input into the compliance pre-validation module for filtering, and after removing non-compliant versions, the final test content set is generated. ,in and To ensure the validity of the test, This represents the number of items retained after filtering by the compliance pre-verification module. Each version of the SMS message content.

5. The SMS sending optimization system based on big data according to claim 4, characterized in that, The specific process by which the generative AI engine generates personalized content candidates is as follows: Template matching: Score users based on their overall characteristics The templates are matched with the applicable feature ranges of the template library, and the three candidate templates with the highest matching degree are selected. Variable population: User features are associated with template variables based on entity linking technology. This entity linking technology performs type parsing on template variable placeholders, establishes a variable type library, and assigns a unique ID to each variable type from the user feature vector. Extract entities that match the type, construct an entity pool, and calculate the semantic association between entities and variables using the following formula: ,in, Candidate entities in the entity pool For template variables, For BERT embedding vectors of entities / variables, To populate the variable with the entity that appears most frequently in the user's historical behavior, select the entity with the highest Rel value. Style Adjustment: Based on the wording style tags in the user tags, the template text is rewritten to adapt to the user and obtain personalized content candidates.

6. The SMS sending optimization system based on big data according to claim 1, characterized in that, The spatiotemporal test matrix construction module includes: Time Set Generation Unit: Used to extract user activity time features from user historical behavior data and generate a set of sending times. ; Matrix building unit: the final set of test content With the set of sending times Combine them to construct a test unit matrix. Each unit Content In time The sending strategy; The output of the spatiotemporal test matrix construction module is a dynamically adjusted test unit matrix. Each unit A corresponding content-time combination strategy is input into the dynamic traffic allocation module for sample allocation.

7. The SMS sending optimization system based on big data according to claim 1, characterized in that, The dynamic traffic allocation module includes: a user grouping unit, a traffic mapping unit, and a traffic balance verification unit, used to allocate user groups to different units of the test unit matrix, wherein: User grouping unit: Based on user profile features, from user feature vectors The target user group was divided into K homogeneous groups based on features highly correlated with SMS response rate. The correlation coefficient of SMS response rate was Pearson correlation coefficient. Filter features, calculate the internal homogeneity and inter-group heterogeneity of each group, and verify the effectiveness of the grouping. The internal homogeneity should satisfy the variance of the SMS response rate of users within the group ≤ 0.1, and the inter-group heterogeneity should satisfy the difference in the mean of the SMS response rate between groups ≥ 0.

2. Traffic mapping unit: Maps users from each group to the test unit matrix at a ratio of 5%. Different units; Balance verification unit: Verifies the consistency between the sample feature distribution of each test unit and the target population, calculates the feature statistics of each test unit sample, and evaluates the degree of difference between the sample feature distribution and the target population. This degree of difference is obtained by calculating the standard deviation of the test unit sample and the target population. Then adjust the traffic allocation ratio and reallocate the samples; The output of the dynamic traffic allocation module is a uniformly distributed test sample, with each test unit... A set of homogeneous user samples is input into the strategy decision feedback module for effect analysis.

Images