Data processing system for advertisement marketing

Abstract

The invention provides a data processing system for advertisement marketing, and relates to the technical field of data query. Advertisement marketing data is acquired through an acquisition module; determining a delivery prediction probability during targeted advertisement delivery through a processing module; the advertisement pushing frequency in advertisement marketing is matched through a matching module; an initial clustering center of multi-source data of advertisement marketing in Internet search is determined through a sorting module, then configuration information for advertisement pushing of target audiences is obtained, and then real-time feedback record tracking tags of users are automatically sorted through the configuration information; and a query module performs distributed query on the index data of consumer preferences during advertisement marketing according to the matched advertisement pushing frequency and the feedback record tracking tag after automatic sorting. According to the method and the device, the distributed query of the index data of consumer preferences during advertisement marketing can be realized under the influence of too heavy sorting load of the distributed query of heterogeneous data; and the accuracy of data query in multi-source data processing is improved.

Description

Technical Field

[0001] The present application relates to the technical field of data query, and more specifically, to a data processing system for advertising marketing. Background Art

[0002] Data query refers to the process of retrieving specific information or data from a database or data storage system. By writing and executing query statements, users can extract the required data from the data set, filter, sort, summarize or calculate to obtain useful information; data query is usually performed in a relational database using structured query language (SQL), or in a non-relational database or distributed data system through a query language or interface; the query results can be used for data analysis, report generation, decision support and other purposes. Effective data query can not only quickly extract and process data, but also ensure the accuracy and relevance of the data, thereby supporting data-driven decision-making and operations.

[0003] In the existing multi-source data processing system for advertising marketing, first, the system collects advertising-related data from different data sources (such as websites, social media, and advertising platforms), including user clicks and feedback records. Then, data integration and preprocessing are performed to ensure data consistency and accuracy. Subsequently, the system extracts key features such as user interests and advertising click-through rates, and builds prediction models based on these features to predict advertising effects and user behaviors. According to the prediction results, the system optimizes advertising strategies, adjusts advertising content and push frequency, and finally, the system monitors advertising effects in real time and adjusts according to user feedback to ensure continuous improvement of advertising strategies. However, in the multi-source data processing of advertising marketing, there are inconsistencies in data formats and structures of different data sources, resulting in uneven data distribution or excessive query load, which leads to excessive sorting load for distributed query of heterogeneous data during multi-source data processing, thereby reducing the accuracy of data query in multi-source data processing. Therefore, how to improve the accuracy of data query in multi-source data processing under the influence of excessive sorting load for distributed query of heterogeneous data is a problem faced by the industry. Summary of the invention

[0004] The present application provides a data processing system for advertising marketing, which can improve the accuracy of data query in multi-source data processing under the influence of excessive sorting load of distributed query on heterogeneous data.

[0005] In a first aspect, the present application provides a data processing system for advertising marketing, the processing system comprising: An acquisition module, used to acquire advertising and marketing data generated during Internet search and stored in the receiving end; A processing module, used to query candidate business data of a target audience group during advertising marketing from the advertising marketing data, and determine a delivery prediction probability during targeted advertising delivery based on the candidate business data; A matching module is used to obtain data status information for personalized recommendation in advertising marketing, determine the advertising push support according to the data status information and multi-source heterogeneous data corresponding to the user's real-time behavior, and then match the advertising push frequency in advertising marketing through the push support and the delivery prediction probability; A sorting module is used to determine the initial clustering center of the multi-source data of advertising marketing in the Internet search, perform parallel clustering on the multi-source data of advertising marketing according to the initial clustering center, obtain configuration information for advertising push to the target audience group, and then automatically sort the user's real-time feedback record tracking tags through the configuration information; The query module is used to perform distributed query on the indicator data of consumer preference in advertising marketing based on the matched advertising push frequency and the automatically sorted feedback record tracking tags.

[0006] In this embodiment, the advertising and marketing data generated during the Internet search and stored in the receiving end is obtained by reading the access log.

[0007] In this embodiment, querying the candidate business data of the target audience group during advertising marketing from the advertising marketing data specifically includes: Extracting query index dimensions for advertising marketing from the advertising marketing data; The advertising and marketing data is filled in according to the query index dimension to obtain candidate business data of the target audience group when querying advertising and marketing.

[0008] In this embodiment, the push distribution rule represents the distribution pattern of advertisement delivery effects in different time periods and different user groups, which is obtained by analyzing historical advertisement delivery data.

[0009] In this embodiment, data status information for personalized recommendation in advertising marketing is obtained by reading the storage unit.

[0010] In this embodiment, data status information refers to information such as the user's current interests, needs, and behavioral tendencies generated by the system after collecting, integrating, and analyzing the user's behavioral data.

[0011] In this embodiment, multi-source heterogeneous data refers to data from different data sources, which are diverse and different in data format, structure, storage method and source.

[0012] In this embodiment, determining the initial cluster center of multi-source data of advertising marketing in Internet search specifically includes: Determine the data sample points corresponding to the multi-source data of advertising and marketing; Determine the cluster density of multi-source data on advertising marketing in Internet searches; The initial cluster center of multi-source data of advertising marketing in Internet search is determined according to the data sample points and the cluster density.

[0013] In this embodiment, the multi-source data of advertising marketing are clustered in parallel according to the initial cluster center to obtain configuration information for pushing advertisements to the target audience group, which specifically includes: Determine the cluster nodes corresponding to the multi-source data of advertising and marketing; Performing proximity analysis on the multi-source data of the advertising marketing through the clustering nodes to obtain data proximity distances; Determining a push distribution balance when pushing advertisements according to the data proximity distance; The configuration information of advertisement push is obtained by integrating the initial cluster center and the push distribution balance.

[0014] In this embodiment, the push distribution balance is an indicator for measuring whether the advertisement pushes are fairly distributed among different cluster nodes.

[0015] The technical solution provided by the embodiments disclosed in this application has the following beneficial effects: The acquisition module acquires the advertising and marketing data generated during Internet search and stored in the receiving end; the processing module queries the candidate business data of the target audience group during advertising and marketing from the advertising and marketing data, and determines the delivery prediction probability during targeted advertising delivery based on the candidate business data; the matching module acquires the data status information for personalized recommendation in advertising and marketing, determines the advertising push support based on the data status information and the multi-source heterogeneous data corresponding to the user's real-time behavior, and then matches the advertising push frequency in advertising and marketing through the push support and the delivery prediction probability; the sorting module determines the initial clustering center of the multi-source data of advertising and marketing in Internet search, and clusters the multi-source data of advertising and marketing in parallel based on the initial clustering center to obtain the configuration information for advertising push to the target audience group, and then automatically sorts the user's real-time feedback record tracking tags based on the configuration information; the query module performs distributed query on the indicator data of consumer preference during advertising and marketing based on the matched advertising push frequency and the automatically sorted feedback record tracking tags.

[0016] It can be seen that in this application, distributed query of multi-source data is implemented under the influence of excessive sorting load of distributed query of heterogeneous data; among them, by collecting advertising and marketing data from multiple sources, the system can establish a comprehensive data set covering users' online behavior and advertising interaction. This comprehensive data capture capability helps to understand user needs and behaviors more accurately; by extracting candidate business data of the target audience from the advertising and marketing data and calculating the delivery prediction probability, the system can accurately identify potential target groups. This precise prediction capability improves the relevance of advertisements, optimizes the efficiency of advertising delivery, and effectively reduces the prediction errors caused by data heterogeneity; by integrating the data status information of personalized recommendations with the real-time behavior of users, the system can calculate the support for advertising push and accurately match the frequency of advertising push, which can improve the personalization of advertisements, ensure that advertising push is highly consistent with user interests, and effectively handle the fusion of multi-source heterogeneous data, reducing the load problem in distributed queries; by performing parallel clustering through the initial clustering center, the system can efficiently group the advertising data to generate accurate advertising push configuration information, improve the efficiency of automatic sorting of user feedback, optimize advertising strategies, reduce the computational burden in multi-source data processing, and improve the accuracy of data processing in distributed queries; through distributed query matching of advertising push frequency and feedback records, the system can quickly process a large amount of consumer preference indicator data. This efficient query capability ensures the accuracy and timeliness of data in the process of multi-source data processing, and effectively deals with the query load problem caused by heterogeneous data, thereby improving the accuracy of advertising marketing decisions and the overall efficiency of the system.

[0017] In summary, the technical solution adopted in the present application can improve the accuracy of data query in multi-source data processing under the influence of excessive sorting load of distributed query on heterogeneous data. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0019] Figure 1 It is a module structure diagram of a multi-source data processing system for advertising and marketing provided by the present application; Figure 2 It is a schematic diagram of a process for determining a delivery prediction probability according to the present application; Figure 3 It is a flowchart of determining the support level of advertising push according to the present application. DETAILED DESCRIPTION

[0020] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0021] The embodiment of the present application provides a data processing system for advertising marketing, the core of which is to obtain advertising marketing data generated during Internet search and stored in a receiving end through an acquisition module; query candidate business data of a target audience group during advertising marketing from the advertising marketing data through a processing module, and determine the delivery prediction probability during targeted advertising delivery based on the candidate business data; obtain data status information for personalized recommendation in advertising marketing through a matching module, determine the advertising push support based on the data status information and multi-source heterogeneous data corresponding to the user's real-time behavior, and then match the advertising push frequency in advertising marketing through the push support and the delivery prediction probability; determine the initial clustering center of the multi-source data of advertising marketing in Internet search through a sorting module, perform parallel clustering on the multi-source data of advertising marketing based on the initial clustering center, obtain configuration information for advertising push to the target audience group, and then automatically sort the user's real-time feedback record tracking tags through the configuration information; and perform distributed query on the indicator data of consumer preference during advertising marketing based on the matched advertising push frequency and the automatically sorted feedback record tracking tags through a query module. The above scheme is adopted to improve the accuracy of data query in multi-source data processing under the influence of excessive sorting load of distributed query on heterogeneous data.

[0022] In order to better understand the above technical solution, the above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods. Figure 1 As shown, this figure is a module structure diagram of a data processing system for advertising marketing according to this embodiment of the present application. The processing system includes: an acquisition module 100, a processing module 200, a matching module 300, a sorting module 400 and a query module 500, which are described as follows: The acquisition module 100 is used to acquire the advertising and marketing data generated during the Internet search and stored in the receiving end.

[0023] In specific implementation, when a user visits a web page, the advertising platform sets a cookie through the HTTP response header. Each subsequent time the user visits the relevant web page again, the browser will automatically send this data to the server. For example, when a user visits an e-commerce website, the website can use cookies to record which products the user has browsed, and automatically generate an access log with the access data corresponding to the access behavior. The advertising marketing data generated when searching the Internet stored in the receiving end can be obtained by reading the access log. In other embodiments, the advertising marketing data generated when searching the Internet stored in the receiving end can also be determined by tracking pixels or third-party data integration, which is not limited here. Cookies are small text files stored on user devices and generated by websites. By storing cookies in user browsers, advertisers can record user behavior data, such as visited pages and clicked ads.

[0024] It should be noted that in this application, advertising and marketing data refers to user behavior, interaction, feedback and other related data collected through various channels and technical means during the process of advertising and marketing activities. These data include but are not limited to users' browsing history, click behavior, number of advertising exposures, purchasing behavior, search keywords, social media interactions, device information and geographic location, etc. Advertising and marketing data is used to analyze users' interests and preferences, help advertisers optimize advertising content, adjust advertising delivery strategies, and make personalized recommendations to improve advertising conversion rates and marketing effectiveness.

[0025] The processing module 200 is used to query the candidate business data of the target audience group during advertising marketing from the advertising marketing data, and determine the delivery prediction probability during targeted advertising delivery based on the candidate business data.

[0026] In this embodiment, querying the candidate business data of the target audience group during advertising marketing from the advertising marketing data can be implemented by the following steps: Extracting query index dimensions for advertising marketing from the advertising marketing data; The advertising and marketing data is filled in according to the query index dimension to obtain candidate business data of the target audience group when querying advertising and marketing.

[0027] In the specific implementation, first, identify the key query dimensions from the advertising and marketing data, such as the user's age, gender, interests, geographic location, device type, etc., and then set the behavioral data dimensions, including the user's search keywords, number of ad clicks, dwell time and other behavioral data. These dimensions can reflect the user's attention to a certain type of advertisement. Then set the historical interaction dimension, such as whether the user has ever purchased a certain type of product or clicked on a certain advertisement. These data are used to predict future behavior; then, the missing data can be inferred based on the known dimensional information and filled in. For example, the user's browsing history may point to a specific interest. The system can infer the category of goods that the user is interested in and fill in the interest dimension. It can also be filled in through statistical methods. For example, when the user's age is missing, the average age of the group can be used to fill it in. The filled dimensional information is reintegrated into the user profile to generate complete candidate business data. This business data contains all the key attributes of the target audience (such as age, interests, click behavior, etc.), and these attributes have been cleaned and improved.

[0028] It should be noted that, in this application, the query index dimension refers to the key data dimension extracted from the advertising and marketing data to describe the characteristics of the target audience group. These dimensions can be the user's social attributes (such as age, gender) or behavioral data (such as search history, number of clicks). The query index dimension is used to help the system quickly screen out the target user groups that meet specific conditions and is the core reference data for advertising delivery; candidate business data refers to the key business data describing the target audience group generated based on the query index dimension and data filling. These data are used to identify potential target customer groups in advertising marketing. The candidate business data is the basis for the advertising system to determine the advertising delivery strategy, including the object, frequency and content of the delivery.

[0029] Preferably, in this embodiment, the delivery prediction probability of targeted advertising delivery is determined according to the candidate business data, referring to Figure 2 As described above, the figure is a schematic diagram of a process for determining a delivery prediction probability in some embodiments of the present application. In this embodiment, determining the delivery prediction probability can be implemented by the following steps: In step S21, a push distribution rule for targeted advertising delivery is determined according to the candidate business data; In step S22, the interest anchor point of the target user is determined according to the push distribution rule; In step S23, a push decision information sequence is generated when pushing advertisements according to the interest anchor point; In step S24, the delivery prediction probability of targeted advertising delivery is determined according to the push decision information sequence.

[0030] In the specific implementation, first, obtain the user's ad click behavior, browsing history, purchase history and other data from the candidate business data, analyze the key indicators such as the click-through rate and conversion rate of the ad push, and then count the frequency and reaction (such as clicks and purchases) of different user groups receiving ads in certain time periods to determine the efficient time periods and areas for delivery, and then obtain the push distribution rules for targeted advertising. Statistical and machine learning algorithms (such as regression analysis, time series analysis, etc.) can be used to analyze historical data to determine the optimal time, frequency and target user groups for ad push. For example, based on historical click data, the system can generate a click probability distribution graph for a certain type of advertisement in different time periods. Then, based on the user's browsing history, purchase history, and ad click behavior, the system can identify the user's preference for specific categories of advertisements. For example, a user clicks on advertisements related to technology products many times, which means that his interest anchors may be concentrated on technology products. Then, the user's interest preferences are extracted through the key dimensions in the candidate business data (such as age, gender, interests, etc.) to generate the user's interest tags. The collaborative filtering, content recommendation algorithm or clustering algorithm (such as K-Means) in the recommendation system can be used to identify the interest anchors in the user's interest tags, and then the interest anchors of the target user can be obtained. For example, the system can compare the user's behavior with that of similar users to infer the user's interest anchors. Then, based on the user's interest anchors, it is matched with the advertising content library to generate recommendations. The sequence of recommended advertisements is combined with the user's interest anchors and the distribution pattern of push notifications to generate decision information for advertising push. For example, the system may decide to push technology product advertisements at a specific time every day. The sorting algorithm (such as gradient boosting decision tree, XGBoost) or rule engine in the recommendation system can be used to generate a push decision information sequence according to the priority of the advertisements. Finally, based on the user's historical behavior and the sequence of advertising push decision information, the delivery prediction model is trained using machine learning models (such as logistic regression, random forest, neural network, etc.), and then the user's interest anchors, advertising content and push strategy are input. The system generates a predicted probability for each advertising delivery. For example, the probability of a user clicking on a technology advertisement may be higher than that on other types of advertisements. A binary classification model can be used to predict the click-through rate (CTR). Common models include logistic regression, support vector machine (SVM), etc., which will not be repeated here.

[0031] It should be noted that in the present application, the push distribution law refers to the distribution pattern of advertising delivery effects in different time periods and different user groups obtained by analyzing historical advertising delivery data. The push distribution law is used to optimize the push frequency, time and target group of advertisements, and improve the delivery effect of advertisements; the interest anchor point refers to the user's high interest point in a certain type of advertisement or product identified based on the user's historical behavior and data characteristics. The interest anchor point is used to determine the content and strategy of advertising push to ensure that the advertisement meets the user's interest needs; the push decision information sequence refers to a series of advertising push instructions generated based on the user's interest anchor point and the push distribution law, which determines the time, frequency, content, etc. of the advertising push. Through this sequence, the system decides how and when to push advertisements to users to maximize the delivery effect; the delivery prediction probability refers to the probability that the advertisement is successfully delivered and clicked by the user, predicted by the advertising system based on user behavior data and push strategy. The delivery prediction probability facilitates a more accurate evaluation of the success rate of advertising delivery and optimization of advertising strategies.

[0032] The matching module 300 is used to obtain data status information for personalized recommendation in advertising marketing, determine the advertising push support based on the data status information and multi-source heterogeneous data corresponding to the user's real-time behavior, and then match the advertising push frequency in advertising marketing through the push support and the delivery prediction probability.

[0033] In specific implementation, the following methods can be used to obtain data status information for personalized recommendations in advertising marketing, namely: first, it is necessary to identify and classify different types of data, including static basic user information (such as basic user information (age, gender, occupation, etc.), geographic location, device information, etc., which will not change frequently but are instructive for personalized recommendations) and dynamic real-time behavior data (user real-time behavior data (such as recent search keywords, browsing history, click records, purchase behavior, etc.) can reflect their current interest changes). Then, by tracking the user's search, click, ad exposure and browsing behaviors, the system can continuously monitor their interest changes. For example, when a user searches for a certain product, the system can record it in real time and convert it into a basis for personalized recommendations. Then, the data received by the system often contains different formats and fields. In order to process these data in a unified manner, the fields in different data sources need to be standardized. For example, to standardize "product clicks" in e-commerce platforms and "post clicks" in social platforms into the same behavior type, you can use ETL (extraction, transformation, loading) tools such as Apache Nifi or Talend is used to standardize and integrate data, and then a big data platform (such as Hadoop or Spark) is used to process and analyze the data. Finally, through the analysis of user behavior data, the system can identify the user's real-time interest preferences, purchasing tendencies and other information. For example, the user's recent frequent searches for travel-related content may indicate that he or she has travel plans. The advertising system can give priority to pushing travel product advertisements. Machine learning algorithms (such as collaborative filtering, content recommendation, deep learning, etc.) can be used to model the user's behavior data, extract their current interests and needs, and use the data after behavior analysis as the data status information for personalized recommendations, which is stored in the storage unit of the large user bureau processing system. The data status information for personalized recommendations in advertising marketing can be obtained by reading the storage unit.

[0034] It should be noted that in this application, data status information refers to the user's current interests, needs, behavioral tendencies, and other information generated by the system after collecting, integrating, and analyzing the user's behavioral data. This information is used to support the advertising system in making personalized recommendation decisions.

[0035] Preferably, in this embodiment, the support for advertising push is determined based on the data state information and the multi-source heterogeneous data corresponding to the user's real-time behavior, referring to Figure 3 The figure is a schematic diagram of a process for determining the support for advertising push in some embodiments of the present application. In this embodiment, determining the support for advertising push can be implemented by the following steps: In step S31, multi-source heterogeneous data corresponding to the user's real-time behavior is obtained; In step S32, the multi-source heterogeneous data is extracted to obtain a push preference fusion amount corresponding to the user's real-time behavior; In step S33, the recommendation decision amount for personalized recommendation is extracted from the data state information; In step S34, the advertisement push support is determined according to the push preference fusion amount and the recommendation decision amount.

[0036] In the specific implementation, first, through various data interfaces and tracking tools, the user's behavior data is collected in real time, including click records, browsing history, search keywords, etc., and the collected user behavior data is used as multi-source heterogeneous data corresponding to the user's real-time behavior. The user behavior data can be obtained from social media, e-commerce websites, etc., which will not be repeated here; then, the data from different sources are sorted and normalized to ensure that their formats are consistent, and then key features are extracted from the sorted data, such as the user's click frequency, browsing time, etc., and the data from different sources are fused and calculated to generate a push preference fusion quantity. For example, the preference information from different sources is integrated into a comprehensive quantity through methods such as weighted average; then, the decision quantity related to personalized recommendation is extracted from the obtained data status information, and it is used as the recommendation decision quantity for personalized recommendation, such as the user's interest tags, recommendation priority, etc., which are not limited here; finally, the push preference quantity and the recommendation decision quantity are combined, and the advertising push support is determined using a weighted method or other comprehensive calculation method. Simple weighted average, or more complex algorithms, such as machine learning models, can be used to comprehensively evaluate the push support.

[0037] It should be noted that in this application, multi-source heterogeneous data refers to data from different data sources, which are diverse and different in data format, structure, storage method and source; the push preference fusion amount is a comprehensive push preference index calculated based on the user's multi-source heterogeneous data, reflecting the user's acceptance of different advertisements; the recommendation decision amount refers to the decision indicator for personalized recommendation extracted from the data status information, including user interest tags, recommendation priority, etc.; advertising push support is an indicator calculated by comprehensively calculating the user's push preference fusion amount and recommendation decision amount, which is used to evaluate the effect of advertising push and the possibility of user acceptance.

[0038] In specific implementation, the following method can be used to match the push frequency of advertisements in advertisement marketing by using the push support and the delivery prediction probability, namely: first, obtain the data of push support and delivery prediction probability, the push support reflects the possibility of advertisement acceptance, and the delivery prediction probability indicates the expected effect of advertisement delivery, and then evaluate the push demand of each advertisement according to the push support and delivery prediction probability. For example, advertisements with high support and high prediction probability may require a higher push frequency; then, formulate frequency matching rules, for example, advertisements with high push support and high prediction probability can increase the push frequency, while advertisements with low support or low prediction probability can reduce the push frequency, and then apply the rules to adjust the actual push frequency of each advertisement. For example, if the push support of an advertisement is 0.8 and the delivery prediction probability is 0.7, its push frequency can be set to a higher value according to the preset rules; finally, use the advertising management system or platform to push advertisements according to the matching rules to ensure that the advertisements are displayed to the target audience at the preset frequency, and monitor the actual performance of the advertisements, including click-through rate, conversion rate, etc., to evaluate the effect of the push frequency and make necessary adjustments.

[0039] It should be noted that, in the present application, the advertisement push frequency refers to the number of times an advertisement is displayed to a user within a certain period of time, and the frequency depends on the push support and delivery prediction probability of the advertisement.

[0040] The sorting module 400 is used to determine the initial clustering center of the multi-source data of advertising marketing in the Internet search, and to perform parallel clustering on the multi-source data of advertising marketing according to the initial clustering center to obtain configuration information for pushing advertisements to the target audience group, and then to automatically sort the user's real-time feedback record tracking tags through the configuration information.

[0041] In this embodiment, determining the initial cluster center of multi-source data of advertising marketing in Internet search can be achieved by using the following steps: Determine the data sample points corresponding to the multi-source data of advertising and marketing; Determine the cluster density of multi-source data on advertising marketing in Internet searches; The initial cluster center of multi-source data of advertising marketing in Internet search is determined according to the data sample points and the cluster density.

[0042] In the specific implementation, first, representative data records are selected from the multi-source data of advertising marketing as data sample points. For example, the user's search records, ad click data, browsing history, etc. can be selected. Each sample point should contain sufficient feature information, such as user interest tags, ad interaction times, etc., so as to be effectively distinguished in cluster analysis; then, the density around each data sample point is calculated to obtain the cluster density of multi-source data in Internet search, that is, the number of data points within a specific radius. For example, the density can be estimated by the number of points within the cluster in the K-means clustering algorithm; finally, data sample points with high cluster density are used as initial cluster centers. Areas with higher density usually contain more relevant data points, which help to form more stable cluster centers. For example, if the cluster density is 0.8, 0.7, 0.6, 0.5, and 0.9, the three points with the highest density are selected as initial cluster centers, that is, 0.9, 0.8, and 0.7, which will not be repeated here.

[0043] It should be noted that in this application, a data sample point refers to a representative single data record extracted from a data set for cluster analysis. The data sample point, as the basic unit of cluster analysis, represents the actual characteristics and distribution of the data; cluster density indicates the concentration of data points in a specific area, usually expressed as the number of data points per unit area, which is used to evaluate the degree of aggregation of data points in the area and help determine the initial cluster center; the initial cluster center refers to the center point selected at the beginning of the algorithm when performing cluster analysis.

[0044] In this embodiment, the multi-source data of advertising marketing are clustered in parallel according to the initial cluster center to obtain configuration information for advertising push to the target audience group, which can be implemented by the following steps: Determine the cluster nodes corresponding to the multi-source data of advertising and marketing; Performing proximity analysis on the multi-source data of the advertising marketing through the clustering nodes to obtain data proximity distances; Determining a push distribution balance when pushing advertisements according to the data proximity distance; The configuration information of advertisement push is obtained by integrating the initial cluster center and the push distribution balance.

[0045] In the specific implementation, first, the initial cluster center is used to allocate the multi-source data sample points of the advertising marketing to the closest cluster node. For example, the K-means algorithm is used to allocate the data to different clusters, and a data point will be marked as the cluster node to which it belongs, that is, its closest initial cluster center; then, for each data sample point, the distance between it and the cluster node to which it belongs is calculated, usually using the Euclidean distance or the Manhattan distance, and the calculation result of the neighboring distance will be used to further analyze the distribution of the data to help evaluate the effect of clustering and the balance of push distribution; then, the distribution of neighboring distances of different cluster nodes is analyzed, and the push load of each node is calculated, and the push load is used as the push distribution balance. In other embodiments, other methods can also be used to determine the push distribution balance, which is not limited here. For example, the average distance of each node can be calculated, and the push frequency can be adjusted accordingly; finally, in combination with the location information of the initial cluster center and the push balance, an advertising push strategy is designed, and the advertising push strategy is used as the configuration information of the advertising push, for example, priority is given to pushing to the cluster node with a higher balance, and then the specific configuration of the advertising push is generated, including the push frequency, target audience allocation, etc., and applied to the advertising system.

[0046] It should be noted that in the present application, the cluster node indicates the specific center point or cluster center to which each data sample is assigned in the cluster analysis, representing the category or group of the data; the data proximity distance indicates the distance between the data sample point and the cluster node to which it belongs, indicating the proximity of the sample point to the cluster center, which is used to evaluate the tightness of the clustering and the balance of the push; the push distribution balance is an indicator to measure whether the distribution of advertising pushes among different cluster nodes is fair; the advertising push configuration information is the advertising push strategy generated by combining the initial cluster center and the push distribution balance, including push frequency, target audience allocation, etc.

[0047] In specific implementation, the automatic sorting of user real-time feedback record tracking tags through the configuration information can be determined in the following manner, namely: first, collect the user's real-time feedback data through the advertising system or related platform, including the time, frequency, page browsing, purchase behavior, etc. of the user clicking on the advertisement, and store the real-time feedback record in the database; then, apply the configuration information to the user feedback data to determine the advertisement configuration to which each feedback record should belong, for example, match the corresponding advertisement configuration based on the user's behavioral characteristics, and then mark the user feedback record according to the configuration information, and use the rule engine or algorithm to mark the feedback record as a tag that meets the specific advertisement push configuration; then, use the machine learning algorithm or rule engine to automatically classify the marked feedback records, for example, the feedback records can be classified into categories such as "high interest", "medium interest", and "low interest", and update the classification results to the system for adjusting the advertisement strategy or push frequency, for example, increasing the advertisement push frequency for "high interest" users; finally, analyze the automatic sorting results, evaluate the effect of the advertisement strategy, for example, evaluate the advertisement click-through rate, conversion rate and other indicators, that is, complete the automatic sorting of the user's real-time feedback record tracking tags.

[0048] It should be noted that in this application, the feedback record tracking tag is a mark used to identify and classify user feedback data on advertising push. The tag is determined based on the user's behavior and feedback characteristics, such as clicks, browsing, conversions, etc., so as to track and analyze the advertising effect; in the process of multi-source data processing, the feedback record tracking tag not only helps data integration and unification, but also improves data quality, supports in-depth data analysis and mining, optimizes advertising strategies, realizes personalized services, and supports real-time feedback and dynamic adjustment.

[0049] The query module 500 is used to perform distributed query on the indicator data of consumer preference during advertising marketing according to the matched advertisement push frequency and the automatically sorted feedback record tracking tags.

[0050] In specific implementation, the distributed query of the indicator data of consumer preference during advertising marketing based on the matched advertising push frequency and the automatically sorted feedback record tracking tags can be implemented in the following way, namely: first, collect the user's feedback records and push frequency data from the advertising system. These data usually include the user's click-through rate, conversion rate, browsing time, etc., and combine the tracking tags of the feedback records with the advertising push frequency data to form a comprehensive data set for query; then, apply the tags of the feedback records to the consumer preference indicator data, for example, apply the "high interest" tag to the user preference data to mark the user's interest in the advertisement, and classify the data according to the advertising push frequency to distinguish the user feedback data under different push frequencies; finally, construct a distributed query statement according to the indicator data to be queried. For example, query the user's average click-through rate, conversion rate, etc., execute the query task on a distributed data processing platform (such as Apache Hadoop, Apache Spark), use the distributed computing capabilities of these platforms to parallel process large-scale data sets, and then analyze the query results. For example, identify the frequency and tag categories of advertisements with high click-through rates and high conversion rates, and adjust the advertisement push strategy based on the analysis results. For example, increase the push frequency for high-interest users and adjust the advertisement content to better match user preferences. Distributed query can efficiently process large-scale data sets and significantly improve the speed and efficiency of data processing. Through the distributed computing platform, detailed indicator data of consumer preferences can be quickly obtained.

[0051] It should be noted that in this application, the indicator data of consumer preferences in advertising marketing refers to various data indicators used to evaluate consumers' interest and interaction with advertising content or products in advertising marketing activities. This indicator reflects consumers' behavior and preferences.

[0052] It can be seen that in this application, distributed query of multi-source data can be achieved under the influence of excessive sorting load of distributed query of heterogeneous data; among them, by collecting advertising and marketing data from multiple sources, the system can establish a comprehensive data set covering users' online behavior and advertising interaction. This comprehensive data capture capability helps to understand user needs and behaviors more accurately; by extracting candidate business data of the target audience from the advertising and marketing data and calculating the delivery prediction probability, the system can accurately identify potential target groups. This precise prediction capability improves the relevance of advertisements, optimizes the efficiency of advertising delivery, and effectively reduces the prediction errors caused by data heterogeneity; by integrating the data status information of personalized recommendations with the real-time behavior of users, the system can calculate the support for advertising push and accurately match the frequency of advertising push, which can improve the personalization of advertisements, ensure that advertising push is highly consistent with user interests, and effectively handle the fusion of multi-source heterogeneous data, reducing the load problem in distributed queries; by performing parallel clustering through the initial clustering center, the system can efficiently group the advertising data to generate accurate advertising push configuration information, improve the efficiency of automatic sorting of user feedback, optimize advertising strategies, reduce the computational burden in multi-source data processing, and improve the accuracy of data processing in distributed queries; through distributed query matching of advertising push frequency and feedback records, the system can quickly process a large amount of consumer preference indicator data. This efficient query capability ensures the accuracy and timeliness of data in the process of multi-source data processing, and effectively deals with the query load problem caused by heterogeneous data, thereby improving the accuracy of advertising marketing decisions and the overall efficiency of the system.

[0053] In summary, the technical solution adopted in the present application can improve the accuracy of data query in multi-source data processing under the influence of excessive sorting load of distributed query on heterogeneous data.

[0054] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems) and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0055] A person skilled in the art may understand that all or part of the steps in the various methods of the above embodiments may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, the storage medium including a read-only memory (ROM), a random access memory (RAM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), a one-time programmable read-only memory (OTPROM), an electronically-erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, magnetic disk storage, magnetic tape storage, or any other computer-readable medium that can be used to carry or store data.

[0056] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

Claims

1. A data processing system for advertising marketing, characterized in that: The processing system comprises: An acquisition module, used to acquire advertising and marketing data generated during Internet search and stored in the receiving end; A processing module, used to query candidate business data of a target audience group during advertising marketing from the advertising marketing data, and determine a delivery prediction probability during targeted advertising delivery based on the candidate business data; A matching module is used to obtain data status information for personalized recommendation in advertising marketing, determine the advertising push support according to the data status information and multi-source heterogeneous data corresponding to the user's real-time behavior, and then match the advertising push frequency in advertising marketing through the push support and the delivery prediction probability; A sorting module is used to determine the initial clustering center of the multi-source data of advertising marketing in the Internet search, perform parallel clustering on the multi-source data of advertising marketing according to the initial clustering center, obtain configuration information for advertising push to the target audience group, and then automatically sort the user's real-time feedback record tracking tags through the configuration information; The query module is used to perform distributed query on the indicator data of consumer preference in advertising marketing based on the matched advertising push frequency and the automatically sorted feedback record tracking tags.

2. A data processing system for advertising marketing as claimed in claim 1, characterized in that: The advertising and marketing data generated during Internet searches stored in the receiving end is obtained by reading the access log.

3. A data processing system for advertising marketing as claimed in claim 1, characterized in that: The candidate business data of the target audience group for advertising marketing is specifically queried from the advertising marketing data: Extracting query index dimensions for advertising marketing from the advertising marketing data; The advertising and marketing data is filled in according to the query index dimension to obtain candidate business data of the target audience group when querying advertising and marketing.

4. A data processing system for advertising marketing as claimed in claim 1, characterized in that: The push distribution law refers to the distribution pattern of advertising delivery effects in different time periods and among different user groups, which is obtained by analyzing historical advertising data.

5. A data processing system for advertising marketing as claimed in claim 1, characterized in that: The data status information for personalized recommendation in advertising marketing is obtained by reading the storage unit.

6. A data processing system for advertising marketing as claimed in claim 1, characterized in that: Data status information refers to the information about the user's current interests, needs, behavioral tendencies, etc. generated by the system after collecting, integrating and analyzing the user's behavioral data.

7. A data processing system for advertising marketing as claimed in claim 1, characterized in that: Multi-source heterogeneous data refers to data from different data sources that are diverse and different in data format, structure, storage method and source.

8. A data processing system for advertising marketing as claimed in claim 1, characterized in that: Determining the initial clustering center of multi-source data of advertising marketing in Internet search specifically includes: Determine the data sample points corresponding to the multi-source data of advertising and marketing; Determine the cluster density of multi-source data on advertising marketing in Internet searches; The initial cluster center of multi-source data of advertising marketing in Internet search is determined according to the data sample points and the cluster density.

9. A data processing system for advertising marketing as claimed in claim 1, characterized in that: The multi-source data of advertising marketing are clustered in parallel according to the initial clustering center to obtain configuration information for advertising push to the target audience group, which specifically includes: Determine the cluster nodes corresponding to the multi-source data of advertising and marketing; Performing proximity analysis on the multi-source data of the advertising marketing through the clustering nodes to obtain data proximity distances; Determining a push distribution balance when pushing advertisements according to the data proximity distance; The configuration information of advertisement push is obtained by integrating the initial cluster center and the push distribution balance.

10. A data processing system for advertising marketing as claimed in claim 1, characterized in that: Push distribution balance is an indicator to measure whether the advertising push is distributed fairly among different cluster nodes.

Images