Campus intelligent consumption analysis system and method based on multi-dimensional data fusion

Abstract

The application discloses a kind of based on multi-dimensional data fusion's campus intelligent consumption analysis system and method, it is related to electric digital data processing technical field.The system includes terminal feature fusion, abnormality identification analysis processing, classification verification and synchronous three big modules.Terminal feature fusion module fuses terminal feature data and time feature data in original multidimensional data, generates behavior correlation feature, reduces data dispersion;Abnormality identification analysis processing module carries out abnormality identification and processing to behavior correlation feature, forms feature data set and verifies validity, improves detection accuracy;Classification verification and synchronous module uploads the classification result after verification as core data set to specified service platform, and synchronizes to its database, improves platform processing efficiency.Through multi-module cooperation, the efficient fusion of campus consumption data, accurate anomaly detection and data synchronization are realized, and the comprehensiveness of campus consumption analysis and service platform processing efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of electronic digital data processing technology, and in particular to a campus intelligent consumption analysis system and method based on multi-dimensional data fusion. Background Technology

[0002] With the advancement of campus digitalization, campus consumption data is experiencing explosive growth, covering multiple scenarios such as canteens, supermarkets, and school buses. Existing technology integrates heterogeneous data from multiple sources, including campus card transaction records, facial recognition payment data, access control system entry and exit times, and library borrowing records, to form a basic dataset containing fields such as consumption timestamps, location, and consumption type (e.g., dining, shopping, utilities). The data is then cleaned, outliers (e.g., negative numbers) are filtered, and missing values ​​are filled. Feature engineering is used to extract derived indicators, such as average daily consumption, consumption frequency during different time periods, consumption proportions in different canteens, and timestamps are broken down into hourly dimensions. Subsequently, machine learning models are used to classify student consumption behavior characteristics, and Long Short-Term Memory (LSTM) networks are used to predict consumption trends. In the anomaly detection stage, dynamic thresholds or the Isolation Forest algorithm are used to identify anomalies such as fraudulent transactions. A sliding time window algorithm is used to analyze fluctuation trends. The system also constructs a visual interface and generates analysis reports containing group characteristics.

[0003] For example, Chinese invention patent CN108647277B discloses a mobile campus integrated service platform and its working method, including: a one-stop service hall module to provide service guides, business consultation, feedback, service appointments, online processing, message push, self-service payment, and data statistics services; a data center to provide data support for the business logic layer and the one-stop service hall module; the business logic layer is responsible for the generation, processing, and transformation of logical data; the business logic layer connects the service layer and the platform layer, classifying school-level business services to various departments according to their business affiliation, with each department processing its own business through asynchronous message passing, and the processing data from each department being integrated through a process center.

[0004] For example, Chinese invention patent CN117708764B discloses an intelligent analysis method for student consumption data based on campus smart cards, which includes: firstly, obtaining the initial parameter values ​​of the model for the overall data points based on the deviation between the overall data points and the regional data points, as well as the fluctuations of the regional data points; secondly, adjusting the initial parameter values ​​of the model based on the adjustment factor of the overall data points to obtain the adjusted model parameters of the overall data points; and thirdly, obtaining the consumption prediction data of all students based on the historical overall consumption time series data and the model parameter values.

[0005] The above-mentioned technology has at least the following technical problems:

[0006] In existing technologies, abnormal consumption data prediction is a core component of ensuring campus consumption safety. Due to the large volume of data and the variety of consumption scenarios and terminal devices involved, which are usually managed by different departments and partners, the data collection standards of multiple terminals are inconsistent, leading to problems such as data loss and format confusion. This results in insufficient accuracy of data matching in time and space, and the fusion data cannot truly reflect the consumption scenario. Consequently, there is a problem of high processing delays for abnormal consumption data during the campus intelligent consumption analysis process. Summary of the Invention

[0007] To address the high processing latency of abnormal consumption data in existing campus intelligent consumption analysis technologies, this invention provides a campus intelligent consumption analysis system and method based on multi-dimensional data fusion. The technical solution is as follows:

[0008] On the one hand, a campus intelligent consumption analysis system based on multidimensional data fusion is provided. This system includes: a terminal feature fusion module, an anomaly identification and analysis processing module, and a classification verification and synchronization module. The terminal feature fusion module is used to fuse the terminal feature data stored in the acquired original multidimensional data with the time feature data to obtain behavioral association features, thereby reducing the dispersion between terminal feature data and time feature data. The anomaly identification and analysis processing module is used to perform anomaly identification analysis based on the acquired behavioral association features to handle anomalies in the behavioral association features. Simultaneously, the anomaly-handled behavioral association features are recorded as a feature dataset and their validity is verified to improve the accuracy of anomaly detection. The classification verification and synchronization module is used to upload the classification results corresponding to the valid feature dataset to a designated service platform as a core dataset, and simultaneously synchronize the core dataset with the database of the designated service platform to improve the processing efficiency of the designated service platform.

[0009] On the other hand, a campus intelligent consumption analysis method based on multidimensional data fusion is provided. The method includes: Step 1, fusing terminal feature data stored in the acquired original multidimensional data with time feature data to obtain behavioral association features; Step 2, performing anomaly identification analysis based on the acquired behavioral association features to handle anomalies in the behavioral association features, and recording the anomaly-handled behavioral association features as a feature dataset and verifying its validity; Step 3, uploading the classification results corresponding to the feature dataset after validity verification as a core dataset to a designated service platform, and simultaneously synchronizing the core dataset with the database of the designated service platform.

[0010] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0011] 1. By using primary key association and time anchoring, this approach effectively solves the problems of inconsistent data collection standards across multiple terminals, insufficient spatiotemporal matching accuracy, and scattered and isolated data. In the primary key association stage, a mechanism for selecting identifier primary keys and matching time similarity is constructed, prioritizing the use of identifier primary keys shared by terminal feature data and time feature data to fundamentally prevent data binding failures caused by inconsistent primary keys. Furthermore, for association pairs with time levels higher than the standard, their precision is reduced to a preset level to ensure consistency. For association pairs with time levels lower than the standard, the time level is upgraded and the identifier is marked to complete, avoiding deviations in behavior scenario reconstruction due to inconsistent time precision. This transforms the originally scattered and formatted terminal and time data into behavioral association features with clear terminal identification, complete time information, and unified precision standards, reducing the dispersion of terminal feature data and time feature data.

[0012] 2. By constructing a quantitative judgment system for the frequency of homologous association and the deviation of association frequency, the problem of ambiguity in the judgment of abnormal behavior is solved. The abnormal frequency index calculated from the two reflects the degree of deviation of the behavioral association characteristics, and the behavioral characteristics are differentiated according to the interval in which the index is located. This avoids misjudgment of high-frequency but stable normal behavior and provides timely warnings of potential security risks. The abnormal frequency analysis of this invention can accurately capture high-risk anomalies such as low-frequency high-fluctuation, and can trigger differentiated prompts for different anomaly types such as high-frequency low-fluctuation to improve the efficiency of anomaly response. Compared with the fuzzy judgment of the prior art, it improves the accuracy and targeting of abnormal behavior identification and provides clear priority guidance for subsequent anomaly handling.

[0013] 3. By dynamically adjusting the processing logic through the feature update cycle, this invention solves the problems of insufficient user-personalized adaptation and inadequate sensitivity to potential security risks in existing technologies for anomaly identification. For behavioral association features marked as potential security risks, the invention calculates the anomaly frequency index deviation by combining it with the currently acquired anomaly frequency index and inputs it into a preset feature update mapping set. This allows for timely capture of potential continuous anomalies, avoiding resource waste caused by excessive updates and ensuring that anomaly identification always keeps pace with actual changes. The difference identification enhancement of this invention accurately captures potential security risks through historical benchmarks and dynamically adjusted update cycles, while avoiding misjudgments of normal behavior. This enhances the ability to identify potential security risks, makes anomaly identification more realistic, and improves the accuracy of risk warnings.

[0014] 4. By employing a multi-level processing logic of cross-validation and amplitude deviation rate verification, the problem of single-dimensional anomaly verification and insufficient reliability of anomaly classification in existing technologies is solved. For potential anomaly associations, slope deviation is calculated; for potential security risks, amplitude deviation is calculated. The multi-level logic of verifying behavioral continuity through trend slope deviation and verifying behavioral mutation through amplitude deviation rate can more comprehensively restore the true state of behavioral characteristics and avoid the limitations of single-dimensional verification. At the same time, by classifying and prioritizing verified and unverified anomalies, high-reliability anomalies can be focused on, reducing invalid checks. This multi-level verification logic improves the reliability and processing efficiency of anomaly classification.

[0015] 5. By employing a payment time supplementation and supplementary labeling mechanism, the problems of low data utilization and incomplete data fusion are solved. When terminal feature data lacks time features, the payment time stored in the terminal feature data is extracted first to obtain the time supplement value. For time feature data that exists but lacks corresponding terminal feature data, no matching is performed to avoid invalid data interfering with the fusion results. The payment time supplementation mechanism of this invention ensures the rationality of the supplemented timestamps and guarantees data transparency through supplementary labeling to avoid misunderstandings in subsequent analysis. At the same time, invalid data that truly lacks terminal features is filtered out to ensure the effectiveness of the fused data and improve data utilization. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 A schematic diagram of the structure of a campus intelligent consumption analysis system based on multi-dimensional data fusion provided in an embodiment of the present invention;

[0018] Figure 2 The flowcharts for the terminal feature fusion module and the anomaly identification and analysis processing module provided in the embodiments of the present invention are as follows:

[0019] Figure 3 A flowchart corresponding to the classification verification and synchronization module provided in this embodiment of the invention;

[0020] Figure 4 A flowchart illustrating a campus intelligent consumption analysis method based on multi-dimensional data fusion, provided as an embodiment of the present invention;

[0021] Figure 5 This is an operation interface for querying behavior-related data of a designated service platform provided in this embodiment of the invention;

[0022] Figure 6 This invention provides an interface for querying meal categories on a designated service platform, as provided in this embodiment. Detailed Implementation

[0023] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0024] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0025] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0026] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0027] This invention provides a campus intelligent consumption analysis system based on multi-dimensional data fusion, such as... Figure 1 The diagram illustrates the structure of a campus intelligent consumption analysis system based on multidimensional data fusion. The system's processing flow can include the following modules: a terminal feature fusion module, an anomaly identification and analysis processing module, and a classification verification and synchronization module. The terminal feature fusion module fuses the terminal feature data stored in the acquired original multidimensional data with time feature data to obtain behavioral association features, thereby reducing the dispersion between terminal feature data and time feature data. The anomaly identification and analysis processing module performs anomaly identification analysis based on the acquired behavioral association features to handle anomalies in the behavioral association features. Simultaneously, it records the anomaly-handled behavioral association features as a feature dataset and performs validity verification to improve the accuracy of anomaly detection. The classification verification and synchronization module uploads the classification results corresponding to the valid feature dataset as a core dataset to a designated service platform, and simultaneously synchronizes the core dataset with the database of the designated service platform to improve the processing efficiency of the designated service platform.

[0028] In this embodiment, the terminal feature fusion module serves as the preliminary foundation, and the behavioral association features it outputs are the core input of the anomaly identification and analysis processing module. After the anomaly identification and analysis processing module completes anomaly processing and validity verification of the behavioral association features, the generated classification results are then used as the processing object of the classification verification and synchronization module. Finally, this module realizes the uploading and synchronization of the core dataset.

[0029] In a specific scenario, such as a campus canteen consumption scenario, the designated service platform can be the Campus Smart Payment platform. After applying the method of this invention, the terminal feature fusion module will fuse the terminal feature data (such as terminal identifier and window) of each POS machine in the canteen with the consumption time feature data, reducing the dispersion of the two types of data, and obtaining behavioral association features containing terminal information corresponding to a user. The terminal information is presented as structured association data, and its presentation format needs to conform to the actual scenario attribute requirements. Here is just an example: the terminal information can be presented as a unique terminal identifier (representing the uniqueness of each terminal: for example, the device serial number is POS-ST-012), the terminal trigger location, and the terminal device attributes, etc. The anomaly identification and analysis processing module performs anomaly identification based on this feature, such as judging whether there is frequent consumption outside of mealtimes, and at the same time performs validity verification to ensure the accuracy of anomaly judgment, forming a feature dataset and classification results. The classification verification and synchronization module then uploads these classification results as the core dataset to the designated service platform and completes synchronization, avoiding the problems of scattered canteen consumption data, inaccurate anomaly detection, and low platform processing efficiency, making campus canteen consumption analysis more efficient and secure.

[0030] like Figure 2 The flowchart shown is a process corresponding to the terminal feature fusion module and the anomaly identification and analysis processing module provided in the embodiment of the invention. Specifically, feature fusion is performed, which is divided into primary key association and time anchoring. Time anchoring is performed after primary key association. The first step of primary key association is divided into feature data quantity verification and missing value verification. The second step of time anchoring is divided into precision reduction alignment and precision completion alignment. After feature fusion is completed, anomaly identification and analysis is performed, which is also divided into two steps: anomaly frequency analysis and difference identification enhancement. The anomaly frequency analysis is further divided into potential anomaly association prompts and potential security risk prompts based on the obtained parameters. The difference identification enhancement is to enhance the ability to identify differences through feature update and adjustment.

[0031] Furthermore, feature fusion includes primary key association to reduce redundancy in the association between terminal feature data and time feature data, and time anchoring to reduce differences in the time dimension, ensuring the alignment of the two types of data in the time dimension. The primary key association involves the following steps: matching the primary keys corresponding to the acquired terminal feature data and time feature data, combined with primary key selection constraints. The primary key selection constraints prioritize primary keys shared by both types of data; if no shared primary key exists, the primary key of the time feature data is converted to a shared primary key through an identifier mapping relationship. Based on the primary key, the terminal-time feature matching result is obtained: if the number of matched time feature data is unique, the corresponding time feature data is directly matched with the terminal feature data; if the number of matched time feature data is not unique, the time feature data with the lowest time similarity is matched first, until all time feature data is matched. Time similarity is represented by the reciprocal of the difference between the feature time of the current behavior and the payment time in the terminal feature data.

[0032] If the terminal feature data has no corresponding time feature data, or the time feature data has no corresponding terminal feature data, the matching is deemed to have failed. Missing value verification is then performed on the terminal feature data that failed to match. Specifically: if the terminal feature data has missing time feature data, the payment time in the terminal feature data is obtained. Simultaneously, based on the payment time and timestamp supplementation relationship (e.g., mapping the payment time to the corresponding timestamp by the minute segment), a time supplementation value is obtained. This time supplementation value is used to fill the timestamps without corresponding time feature data. A set supplementation label is then used to label the time feature data after the corresponding timestamp is filled, completing the missing value verification and terminal-time feature matching. If time feature data exists but there is no corresponding terminal feature data (e.g., the user triggers a payment operation but does not complete the payment, or the terminal does not record payment information), it is marked as invalid terminal feature data, and terminal-time feature matching is not performed. After the terminal-time feature matching is completed, the primary key association result is output, a consumption-time association pair is generated, and time anchoring is performed to reduce the degree of time difference between different types of data.

[0033] Specifically, time anchoring involves the following process: Based on a preset time precision alignment level rule (e.g., the preset level is minute-level), consumption-time association pairs are divided into precision reduction alignment and precision completion alignment. Precision reduction alignment indicates that the current time level of the consumption-time association pair is greater than the preset time precision alignment level, and the current time level is reduced to the preset time precision alignment level. Precision completion alignment indicates that the current time level of the consumption-time association pair is less than the preset time precision alignment level, and based on the current time level deviation and the precision completion mapping set, a time alignment completion value is obtained, and the time level is upgraded. The completed time level is marked with a completion identifier. Based on the time anchoring results, the primary key association results are updated in the time dimension, and the consumption-time association pair with the updated time dimension is used as the final behavioral association feature. If the current time level of the consumption-time association pair is equal to the preset time precision alignment level, no processing is performed, and the currently obtained time level is maintained. The current time level of the current consumption-time association pair is obtained based on the time precision feature of the time field (e.g., second-level, grade-level, hour-level, etc.).

[0034] In this embodiment, the identifier mapping relationship is based on a unified data standard preset by the campus to construct a basic mapping relationship (obtained through a designated service platform, which stores all identifier primary keys corresponding to campus users, such as name, student ID / employee ID, etc.). The mapping relationship is supplemented by using the inherent attributes of the terminal device. Under this mapping relationship, different identifier primary keys can be converted to the same identifier primary key. The acquisition of the time supplement value involves inputting the acquired payment time into a preset timestamp supplement relationship. Through the time association calculation logic built into this supplement relationship, a time supplement value that can be used to fill in the missing time information in the terminal feature data is obtained. The payment time and timestamp supplement relationship is constructed based on the time association between consumption behavior and payment behavior in the campus consumption scenario. It includes mapping logic between payment time and consumption behavior timestamps. This mapping logic presets the payment time directly as the consumption behavior timestamp or the payment time plus or minus the time difference (adapting to actual operation time). The rule for using time precision alignment level as the timestamp of a transaction is usually the subtraction of time difference (e.g., 1-2 seconds for QR code confirmation). The preset time precision alignment level rule is pre-set based on the application scenario of the current terminal device. For example, the precision level of a cafeteria POS machine can be at the second level or at the grade level. The time grade deviation represents the difference between the current time grade and the preset time precision alignment level (e.g., if the current time grade is required to be at the second level, then the deviation is grade one). The precision completion mapping set is based on the preset time precision alignment level rule of the campus to build a basic mapping framework. The mapping rule is optimized by combining the precision deviation pattern of historical terminal time data. The distribution of time precision deviation under different terminal types and different scenarios is statistically analyzed (e.g., in the historical time data of a certain type of POS machine, hour-grade data needs to be upgraded to hour-minute-second, and second-grade data is concentrated in the 00-10 second range). This pattern is incorporated into the precision completion mapping set, and completion is prioritized based on the historical average value of the 00-10 second range.

[0035] By adapting the primary key, even if two types of data do not originally share a common primary key, they can achieve primary key unification through a pre-defined identifier mapping relationship. This ensures that each set of terminal feature data can find a corresponding time feature data anchor, avoiding data isolation caused by incompatible primary keys. For scenarios where multiple sets of time feature data are matched, a filtering logic based on time proximity allows terminal feature data to be bound to the time feature data that best matches its payment time, ensuring the rationality of the association rather than simple random matching. For cases where terminal feature data lacks time information, supplementary time values ​​are generated using the payment time stored by the terminal itself, combined with supplementary rules, and labeled. This preserves valid data that might otherwise be discarded and clarifies data attributes through labels, avoiding misunderstandings about the data source in subsequent analysis. This optimization makes the association between terminals and time data more accurate and comprehensive, providing a high-quality, highly complete data foundation for constructing behavioral association features and reducing subsequent analysis biases caused by data association issues.

[0036] The optimization of the time anchoring process primarily addresses the distortion in scene reconstruction caused by differences in time precision. This provides a unified and reliable time dimension for subsequent anomaly identification, standardizing time data with varying precision to the same level. It eliminates barriers to time dimension differences, ensuring comparability and consistency in the time information of each consumption-time association pair. The optimized time data more accurately recreates the time context of consumption behavior, avoiding time judgment biases caused by precision differences. For example, it prevents the inability to accurately determine whether consumption occurred during non-routine periods due to low precision in some data. This provides more reliable time dimension support for the anomaly identification and analysis module, reducing false positives or false negatives caused by time precision issues and improving the overall reliability of the system's data analysis.

[0037] Furthermore, the anomaly identification analysis includes anomaly frequency analysis to quantify the degree of anomaly in behavioral association features, and differential identification enhancement to improve the ability to identify anomalies in behavioral association features. The anomaly frequency analysis process is as follows: based on the obtained homologous association frequency and association frequency deviation, an anomaly frequency index is obtained to quantify the degree of deviation of behavioral association features. The homologous association frequency represents the total frequency triggered by the terminal within a preset time period, and the association frequency deviation represents the degree of deviation between the homologous association frequency and the preset homologous association frequency. If the obtained anomaly frequency index is less than the minimum value of the preset anomaly frequency index range, it indicates that the fluctuation range of the behavioral association feature is low, and the corresponding behavioral association feature is marked as high-frequency, low-fluctuation data, while a potential anomaly association warning is issued. If the obtained anomaly frequency index is greater than the maximum value of the preset anomaly frequency index range, it indicates that the fluctuation range of the behavioral association feature is large, and the corresponding behavioral association feature is marked as low-frequency, high-fluctuation data, while a potential security risk warning is issued, and a security alert is triggered to indicate the existence of a security risk. If the obtained anomaly frequency index is within the preset anomaly frequency index range, it indicates that the behavioral association feature is currently not abnormal.

[0038] Specifically, the enhanced difference identification involves the following steps: For behavioral association features marked as potential security risks, based on the normal distribution of historical behavioral association features corresponding to low-frequency, high-fluctuation data, and combined with the acquired anomaly frequency index, anomaly frequency index deviation is obtained; the anomaly frequency index deviation is input into the feature update mapping set, and the adjustment value of the feature update cycle of the behavioral association features is obtained through the built-in matching relationship to improve the anomaly identification capability; after completing the feature update cycle adjustment, the corresponding behavioral association features are recorded as the feature dataset.

[0039] The specific expression for the anomaly frequency index I is: In the formula, F represents the frequency of homologous associations, which is the total frequency triggered by the terminal within a preset time period. F0 represents the preset frequency of homologous associations. The first part of the formula represents the deviation standardization term. By dividing the absolute value of the deviation by the baseline frequency, misjudgment caused by the different frequency magnitudes of the terminals themselves is avoided (e.g., small supermarket terminals have low baseline frequencies, so the proportion of small deviations is already high; canteen terminals have high baseline frequencies, so the proportion of large deviations is significant). This ensures that the degree of deviation in different scenarios can be compared horizontally. The latter term σ 2 This represents the variance of the frequency of homologous associations. It is a fluctuation correction term. When the frequency fluctuation is large, this correction term will amplify the indicator, reflecting that the abnormal risk is higher when the deviation is the same but the fluctuation is large.

[0040] In this embodiment, the preset anomaly frequency index range is constructed based on historical behavioral correlation feature data in campus smart consumption scenarios. By collecting historical homologous correlation frequency and correlation frequency deviation data of different consumption scenarios and different terminal devices in the campus, and statistically analyzing the distribution patterns of these data according to scenario classification, the concentrated range of homologous correlation frequency and correlation frequency deviation is determined. The built-in matching relationship is based on historical data in campus consumption scenarios. A feature update mapping set containing the correspondence between the anomaly frequency index deviation range and the feature update cycle adjustment value is pre-constructed. The matching relationship in this mapping set will be combined with the subsequent anomaly verification results of behavioral correlation features corresponding to different anomaly frequency index deviations in history (for example, when the deviation is large, the percentage of shortening the update cycle is obtained based on the deviation, making it easier to capture continuous anomalies; similarly, when the deviation is small, the percentage of extending the cycle is obtained based on the deviation, which can avoid resource waste).

[0041] The optimization of anomaly frequency analysis and enhanced difference identification improves the accuracy and targeting of anomaly identification, providing a reliable foundation for the validity verification of subsequent feature datasets. By constructing a quantitative system of homologous association frequency and association frequency deviation, it is transformed into a hierarchical judgment based on a clear indicator range, and outputs differentiated prompts for different categories. The enhanced difference identification is based on the normal distribution state formed by historical behavior association features, and the feature update cycle is adjusted in a personalized way according to the anomaly frequency indicator deviation. The more suitable update cycle is more sensitive to capture, avoiding the difficulty in timely identification of user-specific anomalies due to a uniform cycle, and also avoiding resource waste. Anomaly frequency analysis completes the initial anomaly classification and risk positioning, while the enhanced difference identification improves the personalized capabilities for high-risk behaviors. Together, they make anomaly identification both quantitatively accurate and flexible, reducing misjudgments and missed judgments caused by ambiguity in traditional detection, improving the system's ability to identify various abnormal behaviors in campus consumption scenarios, and providing high-quality anomaly data support for the generation and uploading of subsequent core datasets.

[0042] like Figure 3The flowchart shown is a classification verification and synchronization module provided in this embodiment of the invention. Specifically, cross-verification is performed by constructing an association feature-time curve. The potential abnormal associations and potential security risks obtained by the previous module are further divided. Potential abnormal associations are divided into unverified anomalies and verified anomalies. Potential security risks are divided into unverified security risks and verified security risks. For verified security risks and verified anomalies, they are uniformly classified as verified and the number of incremental packets transmitted is adjusted. For unverified security risks and unverified anomalies, they are uniformly classified as unverified and the synchronization period is adjusted.

[0043] Further, the validity verification process is as follows: Based on the behavioral association features and corresponding timestamps in the feature dataset, a correlation feature-time curve is constructed with the behavioral association features as the ordinate and time as the abscissa. Cross-validation is then performed based on the correlation feature-time curve. Specifically, for potential abnormal associations, the correlation feature-time curve is divided into segments using a pre-set sliding time window, and the trend slope of each segment is obtained based on linear regression. The difference between the current time window trend slope and the previous time window trend slope is summed and averaged to obtain the time window trend slope deviation. The time window selected here must satisfy the condition of moving from the highest point to the lowest point (or from the lowest point to the highest point, i.e., within one time frame) within a window. There is only one curve slope within the time window. If the current time window is the first time window, the deviation value here is the trend slope value within the first time window. If the obtained trend slope deviation of the time window is within the preset trend deviation range, it indicates that the corresponding behavioral correlation feature is showing a continuous downward trend, and the corresponding potential anomaly is updated to a verified anomaly. If the obtained trend slope deviation of the time window is not within the preset trend deviation range, it indicates that the corresponding behavioral correlation feature is volatile, and the corresponding potential anomaly is updated to a pending verification anomaly. After cross-validation, the verified anomaly and the pending verification anomaly are further verified by preset personnel. The processing priority of the verified anomaly is higher than that of the pending verification anomaly.

[0044] For potential security risks, the amplitude deviation rate is obtained by combining the behavioral correlation features corresponding to the timestamp of the current sliding time window with the average of the behavioral correlation features of a preset number in the previous sliding time window. The specific process is as follows: the preset number of time windows is usually 3 time windows. The arithmetic mean of the behavioral correlation feature data (consumption frequency, etc.) in the current sliding time window and the left and right neighboring windows is calculated to obtain the mean of the behavioral correlation features. If the obtained amplitude deviation rate is within the preset amplitude deviation rate range, it indicates that the behavioral correlation features under the current sliding window are volatile, and the corresponding potential security risk is updated to a security risk to be verified. If the obtained amplitude deviation rate is not within the preset amplitude deviation rate range, it indicates that the terminal feature data under the current sliding window is mutated, and the corresponding potential security risk is updated to a verified security risk to prompt the preset personnel to handle it first. The set of consumption anomaly classification results after the preset personnel have processed the potential abnormal correlations and potential security risks is recorded as the core dataset.

[0045] For verified scenarios, the corresponding core datasets are pushed to the designated service platform database in incremental packets. The number of incremental packets transmitted is adjusted according to the push delay deviation to reduce data verification time. Verified scenarios include verified security risks and verified anomalies. For scenarios to be verified, the corresponding core datasets are transmitted in batches of data packets at a preset synchronization period. The synchronization period is increased according to the transmission duration deviation to improve the transmission frequency of data packets and reduce the data volume of batch data packets. Scenarios to be verified include unverified security risks and unverified anomalies.

[0046] In this embodiment, the pre-set sliding time window is predetermined based on the frequency of terminal triggering devices within a single day. The preset trend deviation interval is represented by the slope deviation concentration range of the trend slope deviation of different potential anomaly associations in each scenario calculated through the sliding time window. This is compared with the average value of a preset number of behavioral association features within the previous sliding time window. Here, the preset number refers to the number of time windows, typically three (which can be adjusted according to the application purpose). The preset amplitude deviation rate interval is obtained similarly to the preset trend deviation interval. The number of incremental packets transmitted is adjusted based on the push delay deviation. This is done by setting a corresponding relationship based on the historical changes in data verification time after adjusting the number of incremental packets under different push delay deviations. When it is necessary to adjust the number of incremental packets, the real-time acquired push delay deviation is input into the "push delay deviation - incremental packet number adjustment value" mapping set to obtain the incremental packet number adjustment value. The "Push Delay Deviation - Incremental Packet Quantity Adjustment Value" mapping set is obtained by analyzing the push delay deviation, the corresponding number of incremental packets transmitted, and the data verification time after transmission when pushing incremental packets to a specified service platform at different times. The correlation between the push delay deviation and the change in verification time after adjusting the incremental packet quantity is analyzed using a learning vector quantization algorithm. Similarly, the process of increasing the synchronization period is similar to the process of adjusting the incremental packet quantity, except that the input deviation is replaced by the transmission duration deviation, and the output becomes the increase adjustment value of the synchronization period. The push delay deviation represents the difference between the obtained push duration and the preset push duration. The preset push duration is obtained by summing and averaging the effective push durations in the current scenario in history. The transmission duration deviation represents the difference between the obtained transmission duration and the preset transmission duration. The preset transmission duration is also represented by the result of summing and averaging the effective values ​​in history.

[0047] By combining scenario-based transmission strategies with dynamic parameter adjustments, the data reception and processing efficiency of the designated service platform was optimized, while ensuring the timely transmission of high-priority data. For verified scenarios, incremental packet-based data transmission avoids redundancy caused by full-volume data transmission. The design of adjusting the number of incremental packets according to the push delay deviation effectively reduces the waiting and computational burden during data verification, allowing the platform to receive and process this type of priority core data more quickly and avoiding delays in high-risk information due to improper transmission methods. For scenarios awaiting verification, batch data packets are transmitted at preset intervals, while the synchronization period is adjusted based on transmission duration deviations. This avoids frequent consumption of transmission resources by non-urgent data and controls the data volume of a single batch packet through periodic adjustments, preventing transmission congestion due to excessively large data packets and ensuring that the platform can stably receive data without affecting other core tasks. This optimization allows data uploads to adapt the transmission method according to the risk level and processing priority of the data, improving the transmission efficiency of high-priority data, rationally allocating transmission resources, reducing the data processing pressure on the platform, and enabling the service platform to more efficiently handle subsequent anomaly handling while avoiding resource waste and data transmission delays.

[0048] like Figure 4 The diagram shows a flowchart of a campus intelligent consumption analysis method based on multidimensional data fusion provided by an embodiment of the present invention. This method includes the following steps: Step 1: merging the terminal feature data stored in the acquired original multidimensional data with the time feature data to obtain behavioral association features; Step 2: performing anomaly identification analysis based on the acquired behavioral association features to handle anomalies in the behavioral association features, and simultaneously recording the anomaly-handled behavioral association features as a feature dataset and verifying its validity; Step 3: uploading the classification results corresponding to the valid feature dataset as a core dataset to a designated service platform, and simultaneously synchronizing the core dataset with the database of the designated service platform.

[0049] In this embodiment, step one fuses terminal feature data and time feature data, reducing the dispersion of the two types of data and solving the problem of insufficient spatiotemporal matching accuracy caused by inconsistent collection standards for multi-source data; step two combines anomaly identification analysis and validity verification, making anomaly judgment more targeted and reducing misjudgments and omissions in traditional anomaly detection; step three uploads synchronized core datasets according to different scenarios, avoiding the platform processing burden caused by unified data transmission, improving the efficiency of designated service platforms, and effectively improving the problems of high latency in processing abnormal consumption data and difficulty in reflecting real consumption scenarios in existing technologies.

[0050] like Figure 5As shown, this is the operation interface for querying behavior-related data of a designated service platform provided in this embodiment of the invention. The interface uses a multi-section layout to intuitively display multi-dimensional data such as consumption, usage by time period, total number of users (students, faculty and staff) and consumption distribution, daily consumption trend, consumption frequency at different times, and number of transactions. At the same time, there are function navigation entrances at the bottom such as detailed query, card and coupon management, and user management, which facilitates a comprehensive understanding and management of various information related to campus consumption. This interface achieves real-time aggregation and intelligent presentation of campus consumption data through the integrated visualization of multi-dimensional data and the efficient integration of functional modules. It strengthens the platform's ability to track and analyze consumption behavior across the entire chain, optimizes data interaction response efficiency, provides accurate data support for platform operation decisions and function iteration, and effectively improves its service efficiency and ecosystem integration capabilities in the campus smart consumption scenario.

[0051] like Figure 6 The diagram shows the interface for querying meal categories on a designated service platform provided in this embodiment of the invention. The left side features a navigation bar with entry points for multiple modules, including temperature rule settings and the consumption system. The right side, in its core area, clearly lists the meal category names, start times, normal dining times (exceeding the time limit is defined as abnormal dining time), and normal dining consumption (exceeding the time limit is defined as abnormal dining time) for breakfast, lunch, dinner, late-night snacks, and extra-late-night snacks. The bottom includes operation prompts and editing buttons. This interface achieves full-time coverage and precise control of campus dining behavior through refined rule settings for meal categories, dining times, and other consumption patterns. Clearly defined normal dining times and consumption ranges provide a quantitative benchmark for identifying abnormal consumption, helping canteen operators to accurately optimize meal service rhythms and resource allocation. It also lays a solid rule foundation for abnormal consumption warnings and consumption data statistical analysis, promoting the standardization and intelligentization of campus consumption management and significantly improving operational efficiency and service accuracy.

[0052] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0053] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0054] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0055] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0056] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0057] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the devices, apparatuses, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0058] In the embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0059] The units described as separate components may or may not be physically separate. 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 units can be selected to achieve the purpose of this embodiment according to actual needs.

[0060] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0061] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0062] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A campus intelligent consumption analysis system based on multi-dimensional data fusion, characterized in that, The application relates to a terminal feature fusion module, an abnormality identification analysis processing module and a classification verification and synchronization module. The terminal feature fusion module is used for performing feature fusion on terminal feature data stored in acquired original multidimensional data and time feature data, so as to obtain behavior correlation features, reduce the dispersion degree between the terminal feature data and the time feature data, and obtain the terminal feature data, which comprises a terminal identifier and a corresponding window. The abnormality identification analysis processing module is used for performing abnormality identification analysis on the acquired behavior correlation features, performing abnormality processing on the behavior correlation features, recording the behavior correlation features after the abnormality processing as a feature data set and performing effectiveness verification, so as to improve the accuracy of abnormality detection. The classification verification and synchronization module is used for uploading the classification results corresponding to the feature data set after the effectiveness verification to a specified service platform as a core data set, and synchronizing the core data set with a database of the specified service platform, so as to improve the processing efficiency of the specified service platform. The feature fusion comprises primary key association for reducing the correlation redundancy of the terminal feature data and the time feature data, and time anchoring for reducing the time dimension difference. The primary key association comprises the following steps: An identifier primary key corresponding to the terminal feature data and the time feature data is acquired, and an identifier primary key selection constraint is matched. The identifier primary key selection constraint indicates that a shared identifier primary key of the two types of data is preferentially selected, and if there is no shared primary key, a representation primary key of the time feature data is converted into a shared primary key through an identifier mapping relationship. Based on the identifier primary key, a terminal-time feature matching result is obtained. If the number of the matched time feature data is unique, the corresponding time feature data is directly matched with the terminal feature data. If the number of the matched time feature data is not unique, the time feature data with the largest time proximity degree is preferentially matched, until all the time feature data are matched, and the time proximity degree is represented by the result of the inverse processing of the difference between the time stamp recorded by the current behavior time feature and the payment time in the terminal feature data. The effectiveness verification comprises the following steps: Based on the behavior correlation features and the corresponding time stamps in the feature data set, an association feature-time curve is constructed, with the behavior correlation features as the ordinate and the time as the abscissa. According to the association feature-time curve, cross verification is performed, and the specific steps are as follows: For potential abnormal association, the association feature-time curve is segmented by a pre-set sliding time window, and each segment obtains a time window trend slope based on linear regression. Based on the difference between the current time window trend slope and the last time window trend slope, a time window trend slope deviation is obtained by summation and averaging. If the acquired time window trend slope deviation is within a pre-set trend deviation interval, it is indicated that the corresponding behavior correlation features show a continuous downward trend, and the corresponding potential abnormal association is updated as a verified abnormality. If the acquired time window trend slope deviation is not within the pre-set trend deviation interval, it is indicated that the corresponding behavior correlation features have volatility, and the corresponding potential abnormal association is updated as a to-be-verified abnormality. ​ The verified exception and the to-be-verified exception are verified by preset personnel after cross-validation, and the processing priority of the verified exception is higher than that of the to-be-verified exception.

2. The campus intelligent consumption analysis system based on multi-dimensional data fusion according to claim 1, wherein, The primary key association further comprises: If the terminal feature data has no corresponding time feature data, or the time feature data has no corresponding terminal feature data, it is determined that the matching fails, and missing value verification is performed on the terminal feature data that fails to match, specifically: If the terminal feature data has missing time feature data, the payment time in the terminal feature data is obtained, and based on the payment time and the timestamp supplement relationship, a time supplement value is obtained to fill in the timestamp of the terminal feature data that has no corresponding time feature data, and the time feature data after the corresponding timestamp is filled in is labeled with a set supplement label to complete the missing value verification and terminal-time feature matching; If there is time feature data but no corresponding terminal feature data, it is marked as invalid terminal feature data, and no terminal-time feature matching is performed; After the terminal-time feature matching is completed, the primary key association result is output, the consumption-time association pair is generated, and time anchoring is performed to reduce the time difference between different types of data.

3. The campus intelligent consumption analysis system based on multi-dimensional data fusion according to claim 1, characterized in that, The time anchoring specifically comprises: Based on a preset time precision alignment level rule, the consumption-time association pair is divided into precision reduction alignment and precision complete alignment; The precision reduction alignment indicates that the current time level of the consumption-time association pair is greater than the preset time precision alignment level, and the current time level is reduced to the preset time precision alignment level; The precision complete alignment indicates that the current time level of the consumption-time association pair is less than the preset time precision alignment level, and based on the current time level deviation and the precision complete mapping set, a time alignment complete value is obtained to upgrade the time level, and the complete time level is marked with a complete identifier; If the current time level of the consumption-time association pair is equal to the preset time precision alignment level, no processing is performed, and the current obtained time level is maintained; Based on the time anchoring result, the time dimension of the primary key association result is updated, and the consumption-time association pair after the time dimension update is taken as the final behavior association feature.

4. The campus intelligent consumption analysis system based on multi-dimensional data fusion of claim 1, wherein, The abnormality recognition analysis comprises abnormal frequency analysis for quantifying the abnormality degree of the behavior association feature, and difference recognition enhancement for enhancing the ability to recognize the abnormality of the behavior association feature; The abnormal frequency analysis specifically comprises: Based on the obtained homologous association frequency and association frequency deviation, an abnormal frequency index for quantifying the deviation degree of the behavior association feature is obtained, the homologous association frequency represents the total frequency of terminal triggering in a preset time period, and the association frequency deviation represents the deviation degree of the homologous association frequency from a preset homologous association frequency; If the obtained abnormal frequency index is less than a preset minimum value of an abnormal frequency index interval, it indicates that the fluctuation amplitude of the behavior association feature is low, and the corresponding behavior association feature is marked as high-frequency low-fluctuation data, and a potential abnormality association prompt is performed; If there is time feature data but no corresponding terminal feature data, it is marked as invalid terminal feature data, and no terminal-time feature matching is performed; After the terminal-time feature matching is completed, the primary key association result is output, the consumption-time association pair is generated, and time anchoring is performed to reduce the time difference between different types of data. The time anchoring specifically comprises: Based on a preset time precision alignment level rule, the consumption-time association pair is divided into precision reduction alignment and precision complete alignment; The precision reduction alignment indicates that the current time level of the consumption-time association pair is greater than the preset time precision alignment level, and the current time level is reduced to the preset time precision alignment level; The precision complete alignment indicates that the current time level of the consumption-time association pair is less than the preset time precision alignment level, and based on the current time level deviation and the precision complete mapping set, a time alignment complete value is obtained to upgrade the time level, and the complete time level is marked with a complete identifier; If the current time level of the consumption-time association pair is equal to the preset time precision alignment level, no processing is performed, and the current obtained time level is maintained; Based on the time anchoring result, the time dimension of the primary key association result is updated, and the consumption-time association pair after the time dimension update is taken as the final behavior association feature. The abnormality recognition analysis comprises abnormal frequency analysis for quantifying the abnormality degree of the behavior association feature, and difference recognition enhancement for enhancing the ability to recognize the abnormality of the behavior association feature; The abnormal frequency analysis specifically comprises: Based on the obtained homologous association frequency and association frequency deviation, an abnormal frequency index for quantifying the deviation degree of the behavior association feature is obtained, the homologous association frequency represents the total frequency of terminal triggering in a preset time period, and the association frequency deviation represents the deviation degree of the homologous association frequency from a preset homologous association frequency; If the obtained abnormal frequency index is less than a preset minimum value of an abnormal frequency index interval, it indicates that the fluctuation amplitude of the behavior association feature is low, and the corresponding behavior association feature is marked as high-frequency low-fluctuation data, and a potential abnormality association prompt is performed; If the obtained abnormal frequency index is greater than the preset maximum value of the abnormal frequency index interval, it is indicated that the fluctuation amplitude of the behavior correlation feature is large, and the corresponding behavior correlation feature is marked as low-frequency high-fluctuation data, and a potential security risk prompt is performed, and a security warning is triggered to prompt the existence of a security risk; If the obtained abnormal frequency index is within the preset abnormal frequency index interval, it is indicated that the behavior correlation feature is not abnormal.

5. The campus intelligent consumption analysis system based on multi-dimensional data fusion according to claim 4, characterized in that, The difference recognition reinforcement specifically includes the following steps: For the low-frequency high-fluctuation data, based on the normal distribution state of the corresponding historical behavior correlation feature, and combined with the obtained abnormal frequency index, the abnormal frequency index deviation is obtained; The abnormal frequency index deviation is input into the feature update mapping set, and the adjustment value of the feature update period of the behavior correlation feature is obtained through the built-in matching relationship, so as to improve the abnormal recognition ability; After completing the feature update period adjustment, the corresponding behavior correlation feature is recorded as a feature data set.

6. The campus intelligent consumption analysis system based on multi-dimensional data fusion according to claim 5, characterized in that, The effectiveness verification further includes: For the potential security risk, the behavior correlation feature corresponding to the current sliding time window time stamp is combined with the mean value of a preset number of behavior correlation features in the previous sliding time window to obtain a magnitude deviation rate; If the obtained magnitude deviation rate is within the preset magnitude deviation rate interval, it is indicated that the behavior correlation feature under the current sliding window exists fluctuation, and the corresponding potential security risk is updated to a to-be-verified security risk; If the obtained magnitude deviation rate is not within the preset magnitude deviation rate interval, it is indicated that the terminal feature data under the current sliding window exists mutation, and the corresponding potential security risk is updated to a verified security risk, prompting a preset person to prioritize processing; After the preset person processes the potential abnormal correlation and the potential security risk, the set of consumption anomaly classification results corresponding to the core data set is recorded as a core data set.

7. The campus intelligent consumption analysis system based on multi-dimensional data fusion according to claim 1, characterized in that, The uploading to the designated service platform specifically includes the following steps: For the verified scene, the corresponding core data set is pushed to the designated service platform database in the form of an incremental package, and the number of incremental packages is adjusted according to the push delay deviation to reduce the data verification time consumption, and the verified scene includes a verified security risk and a verified anomaly; For the to-be-verified scene, the corresponding core data set is transmitted in the form of a batch data package according to a preset synchronization period, and the synchronization period is increased according to the transmission time length deviation to increase the transmission frequency of the data package and reduce the data amount of the batch data package, and the to-be-verified scene includes a to-be-verified security risk and a to-be-verified anomaly.

8. The method applied to the campus intelligent consumption analysis system based on multi-dimensional data fusion according to any one of claims 1-7, characterized in that, The method includes the following steps: Step one, performing feature fusion on the terminal feature data and the time feature data stored in the obtained original multi-dimensional data to obtain a behavior correlation feature; Step two, performing abnormal recognition analysis on the obtained behavior correlation feature to perform abnormal processing on the behavior correlation feature, and recording the behavior correlation feature after the abnormal processing as a feature data set and performing effectiveness verification; Step three, uploading the classification results corresponding to the feature data set after the effectiveness verification to the designated service platform as a core data set, and synchronizing the core data set with the designated service platform database.

Images