Cross-platform advertisement putting and SEO collaborative optimization system

Abstract

The invention relates to the technical field of data management, in particular to a cross-platform advertisement putting and SEO collaborative optimization system which comprises a resource time sequence scheduling module, a keyword screening module, a putting path construction module, a data conflict comparison module and a content archiving and auditing module. According to the invention, through association mining of multi-platform advertisement data and user behaviors, real-time mapping of time sequence distribution of put resources and user group states, and dynamic integration of budget weights and audience portraits in a keyword selection process, a multi-parameter interactive screening mechanism is formed for advertisement paths; text similarity judgment and conflict source tracing are synchronously introduced into channel content, an archiving standard is jointly defined by content change frequency and source credibility, data in each link automatically adjusts an attribution boundary according to a change trend, multi-source data validity is accurately distinguished in an archiving process, and structure-level linkage is realized by advertisement circulation, content integration, path adaptation and conflict detection. And the consistency and uniqueness of data updating and archiving are guaranteed.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data management, in particular to a cross-platform advertisement delivery and SEO collaborative optimization system. BACKGROUND

[0002] The technical field of data management relates to the collection, storage, retrieval, transmission and utilization of information, covering the ordered organization, classified management, effective query, real-time update, integration analysis and security protection of data, and can provide basic data support services for various information systems, business decisions, resource scheduling and business processes, and realize the orderly flow and efficient control of data from generation to application. Among them, the traditional cross-platform advertisement delivery and SEO collaborative optimization system refers to the specific measures of advertisement creative content generation, delivery plan formulation, advertisement position allocation, keyword configuration, budget allocation, delivery channel screening, and content optimization, keyword layout, site structure optimization, external link construction, and traffic monitoring in SEO during the process of advertisement delivery and search engine optimization on multiple digital platforms. Through the separate optimization and distribution of advertisement materials and SEO content, the collaborative management of advertisement display and natural traffic acquisition on each platform is realized.

[0003] In the prior art, each advertisement and SEO optimization operation is mainly based on separate templates and platforms, resource division relies on static classification and manual planning, real-time user state mapping is lacking in the actual delivery stage, budget and active behavior are difficult to consider simultaneously in the keyword configuration process, advertisement path matching is limited by channel segmentation, data synchronization often cannot timely discover conflicts due to differences in text specifications, and content archiving lacks weight screening based on credibility and change frequency, resulting in data redundancy, audience distribution deviation and archiving coverage omission, affecting the efficiency of advertisement distribution and the stability of traffic integration. SUMMARY

[0004] The purpose of the present application is to solve the shortcomings in the prior art, and a cross-platform advertisement delivery and SEO collaborative optimization system is proposed.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solution: a cross-platform advertisement delivery and SEO collaborative optimization system, the system comprising: A resource time sequence scheduling module analyzes the relationship between resource parameters, user coverage and time nodes based on an advertisement delivery plan form, compares the time overlap intervals of each resource, analyzes the overlap of traffic distribution nodes and user intervals, adjusts the order of resource time nodes, and obtains resource delivery overlap characteristics; A keyword screening module analyzes product classification tags, audience portraits and user active time periods based on the resource delivery overlap characteristics, compares the click frequency of the word and the budget, judges the relationship between the active time of the keyword and the overlap area, screens the word with the same active time and sorts them according to the click frequency, and obtains a keyword sorting tag set. The resource putting coincident feature includes time sequence distribution characteristics, audience coincident characteristics and putting structure characteristics, the keyword sorting label set includes label identification, preferred level and active period, the path user coverage distribution includes channel distribution characteristics, user quantity distribution and putting time period characteristics, and the abnormal field conflict index includes abnormal identification, conflict type and processing mark. The data conflict comparison module judges the matching degree of similar fields based on the path user coverage distribution, filters the fields with a matching proportion that does not meet the standard, gives a to-be-reviewed mark, optimizes abnormal data detection, and obtains an abnormal field conflict index.

[0006] The resource putting coincident feature includes time sequence distribution characteristics, audience coincident characteristics and putting structure characteristics, the keyword sorting label set includes label identification, preferred level and active period, the path user coverage distribution includes channel distribution characteristics, user quantity distribution and putting time period characteristics, and the abnormal field conflict index includes abnormal identification, conflict type and processing mark.

[0007] The resource time sequence scheduling module includes: The resource coincident analysis submodule analyzes resource attributes and user distribution based on an advertisement putting plan form, judges the intersection of the region and interest label of the user group of each putting channel, compares the coincidence degree of the resource putting time points, calculates the repeated contact frequency of the intersection group on the difference platform, and obtains a resource cross-coverage proportion. The flow interval comparison submodule analyzes the matching situation of flow nodes and audience labels based on the resource cross-coverage proportion, compares the overlap of user attention labels and resource flows in different time periods, and obtains a label overlap density factor by counting the distribution of high-frequency labels in active flow intervals. The time sequence adjustment submodule analyzes the marginal gain change and display frequency of each resource based on the label overlap density factor, compares the distribution difference of features and time sequences, obtains a time structure change amount, adjusts the time node sequence of the resource, and obtains a resource putting coincident feature.

[0008] The keyword screening module includes: The portrait active extraction submodule identifies user portrait information under product labels based on the resource putting coincident feature, analyzes the access frequency of user groups in each time period, judges the access time period distribution of various types of users, determines the time period of the access peak and the active set, and obtains a user active time period proportion. The keyword coincident comparison submodule filters the click behavior of keywords in different time periods according to the user active time period proportion, compares the coincidence of keyword active time periods and advertisement putting time nodes, judges the overlapping range of the coincident time period, filters keywords with a high time coincidence degree, and obtains a keyword time coincidence breadth. The click frequency ranking sub-module calls the keyword time overlap range, optimizes the budget configuration and click performance corresponding to the keyword, compares the click efficiency and user coverage of the unit budget, calculates the ranking factor of each keyword, arranges the keywords in order of ranking, and obtains a keyword ranking label set.

[0009] The application improves that the delivery path construction module comprises: The audience attribute analysis submodule identifies the associated audience portrait features and product attribute content based on the keyword ranking label set, analyzes the correspondence of gender, age and interest audience labels and product categories, compares the population distribution under the difference labels and the keyword combination, and obtains label population distribution information. The resource matching screening submodule analyzes the matching degree of product labels and keyword combinations of each resource node according to the label population distribution information, judges the time overlap of keyword active period and resource delivery window, screens resource nodes with consistent labels and sufficient time intersection, and obtains a node label overlap list. The user coverage statistics submodule obtains target user group information according to the node label overlap list, analyzes user behavior data in the keyword active period, classifies and compares the number of user groups under each channel, time period and region, and obtains path user coverage distribution.

[0010] The application improves that the data conflict comparison module comprises: The text matching comparison submodule analyzes the text content of each channel synchronization field based on the path user coverage distribution, compares the similarity of character content of different sources under the same field category, judges the matching difference of each group of field content, calculates the similarity interval of each group of fields, and obtains the field matching deviation degree. The conflict field screening submodule screens the fields that do not meet the unified comparison standard based on the field matching deviation degree, judges the source, field category and content thereof, analyzes the field distribution and state corresponding to the screening result, sets the screening field as a to-be-reviewed mark, and obtains a to-be-reviewed field set. The abnormal index acquisition submodule analyzes the source, generation time and modification frequency of the field according to the to-be-reviewed field set, compares the number of sources under the current review state, acquires the field conflict offset, screens the abnormal features of the difference type field, and obtains an abnormal field conflict index.

[0011] The application improves that the system further comprises: The content archiving and auditing module analyzes the source channel, generation time and modification frequency of the to-be-reviewed field based on the abnormal field conflict index, identifies the trusted field as a standard, compares the similarity of field content of each channel, merges the fields meeting the standard, and statistically obtains the proportion of merged and archived fields, and obtains an archived content coverage proportion. The archiving content coverage ratio comprises a merging proportion, a credible proportion and an archiving range.

[0012] The content archiving auditing module comprises: The credible field identification submodule analyzes the source channel of the field to be reviewed based on the abnormal field conflict indicator, compares the generation time of each field with the historical modification record, judges whether the field meets the requirements of coming from a specified channel and having a concentrated generation time and a distributed modification frequency, and obtains a credible field screening set; The similarity merging judgment submodule analyzes the character structure of the credible field screening set and the character structure of the fields from each channel based on the credible field screening set, compares the differences between the contents, judges whether the character contents meet the consistency requirements of the merging benchmark, merges the fields meeting the conditions, and obtains a merging field comparison result; The archiving proportion statistical submodule counts the number of the merged fields in the overall fields according to the merging field comparison result, analyzes the proportion of the fields from the credible field, judges the coverage range of the merged fields and the credible fields, and obtains the archiving content coverage ratio.

[0013] Compared with the prior art, the advantages and positive effects of the present application are that: In the present application, the time sequence distribution of the resource distribution and the state of the user group are mapped in real time through the association mining of the multi-platform advertising data and the user behavior, the keyword selection process dynamically integrates the budget weight and the audience portrait, a multi-parameter interactive screening mechanism is formed for the advertising path, the text similarity judgment and the conflict source tracing are synchronously introduced for the channel content, the archiving standard is jointly defined by the content change frequency and the source credibility, the data of each link is automatically adjusted to the attribution boundary according to the change trend, the effectiveness of the multi-source data is accurately distinguished in the archiving process, the structure cascade is realized for the advertising circulation, the content integration, the path adaptation and the conflict detection, and the data update and the archiving guarantee the consistency and the uniqueness. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 The system flowchart of the present application is shown in the figure; Figure 2 The flowchart of the resource time sequence scheduling module in the present application is shown in the figure; Figure 3 The flowchart of the keyword screening module in the present application is shown in the figure; Figure 4 The flowchart of the advertising path construction module in the present application is shown in the figure; Figure 5 The flowchart of the data conflict comparison module in the present application is shown in the figure; Figure 6 The flowchart of the content archiving auditing module in the present application is shown in the figure. DETAILED DESCRIPTION

[0015] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0016] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship shown in the drawings, which are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0017] Embodiments Please refer to Figure 1 The present application provides a technical solution: a cross-platform advertisement delivery and SEO cooperative optimization system comprising: The resource time sequence scheduling module analyzes the correspondence between resource parameters, user coverage and delivery time nodes based on the advertisement delivery plan form, judges the distribution of platform identifiers and user groups in each regional group, compares the time node overlapping intervals of each resource, analyzes the overlapping sections of traffic distribution nodes and user intervals, calculates the difference between marginal gain changes and resource delivery intensity, adjusts the time node order of the resource, and obtains the resource delivery overlapping characteristics; The keyword screening module analyzes the corresponding audience portrait and user active time period of the product classification label based on the resource delivery overlapping characteristics, compares the click frequency of each keyword with the budget data, judges the correspondence between the active time of the keyword and the overlapping area, screens the keywords with consistent active time and overlapping area, and optimizes the keyword sorting according to the click frequency, to obtain a keyword sorting label set; The delivery path construction module analyzes the corresponding audience portrait parameters and product attribute parameters based on the keyword sorting label set, optimizes the matching relationship between platform resources, target audience groups and delivery windows in combination with the influence of the budget, screens resource nodes with overlapping attributes, compares the correspondence between keyword active time and delivery windows, and classifies and organizes the number of target user groups, to obtain path user coverage distribution; The data conflict comparison module compares the text content of the synchronous fields of each channel based on the path user coverage distribution, judges the character matching degree between similar field texts, screens fields with a character matching ratio that does not meet the standard, assigns a to-be-reviewed mark, and optimizes the abnormal data detection process in combination with the field source channel, generation time and modification record, to obtain an abnormal field conflict index; The content archiving review module analyzes the source channel information, generation time and modification frequency of the field to be reviewed based on the abnormal field conflict index, identifies the trusted field with optimal section of modification frequency and generation time as the standard, compares the similarity distribution of the field content of each channel, merges the fields with similarity meeting the synchronization standard, and counts the proportion distribution of the merged and archived fields to obtain the archiving content coverage proportion.

[0018] The resource delivery coincidence feature includes time sequence distribution characteristics, audience coincidence characteristics, and delivery structure characteristics. The keyword sorting label set includes label identification, preferred level, and active period. The path user coverage distribution includes channel distribution characteristics, user number distribution, and delivery time period characteristics. The abnormal field conflict index includes abnormal identification, conflict type, and processing mark. The archiving content coverage proportion includes merging proportion, trusted proportion, and archiving range.

[0019] In module 1, the resource parameter refers to specific ad slots, ad types, display positions, budget allocation, and other attribute information directly related to ad resources in the ad delivery process. User coverage rate refers to the proportion of users that a certain ad resource can reach or influence within a specific platform and time period, and is usually used to measure ad reach effect. Platform identification refers to a unique label that distinguishes ad delivery channels (such as Tencent ads, Baidu information flow, Douyin, and Toutiao). It is used to attribute resource and user data. Distribution in each area group refers to the group division of ad resources based on user attributes (such as region, interest, age, and gender), as well as the delivery coverage of resources in each sub-group. Time node overlap interval refers to the overlapping section of different ad resources in the delivery schedule, which is used to analyze whether there is resource duplication. User interval refers to the user subset formed after grouping according to the user portrait, such as the "18-24 year-old, interest in sports, and users in first-tier cities" set targeted by a certain ad resource. Marginal gain change refers to the change trend of new ad effects (such as new exposure, clicks, and conversions) brought by each unit of resource investment, which is used to judge the cost-effectiveness of continued investment. Resource delivery intensity refers to the density of ad delivery per unit of time or budget, such as the ad display frequency or ad budget consumption rate in a certain time period. Time node sequence refers to the arrangement order of each ad resource in different time, which is used to optimize the overall ad delivery timing and resource utilization efficiency by adjusting the order.

[0020] In module 2, the product classification label refers to a label for classifying the promoted product, such as "sports shoes", "makeup and skin care", "household appliances", etc., which facilitates the association of keywords with specific products; the audience portrait refers to a combination of user characteristics formed based on user behavior data, interest labels, demographic attributes, etc., such as "male, 25-35 years old, and likes technology and digital"; the keyword active time refers to the main active period of a keyword in a certain period of time, which is searched, clicked, and focused on, and is used to select the best delivery time point; the overlap area refers to the intersection of the keyword active time and the actual delivery time segment of the advertising resource, which is one of the core bases for keyword filtering; the optimized sorting term refers to the keyword sequence obtained after prioritizing the filtered keywords according to their click frequency, budget matching degree, etc.

[0021] In module 3, the product attribute parameter refers to a parameter that describes the characteristics of the advertising product itself, such as brand, model, price range, and applicable scenario information, which is used to match with audience preferences; the platform resource refers to all advertising positions, display channels, and corresponding resource pool information that can be deployed and used on various advertising platforms; the delivery window refers to the time period or cycle that allows the advertisement to go online, such as a certain advertisement can only be delivered between 18:00-22:00 every day; the attribute-coinciding resource node refers to an advertising position or resource unit in the platform resource that has complete (or high) consistency in product attributes, audience portrait, and budget elements.

[0022] In module 4, the text content of the synchronization field refers to the text content of the same type of data field (such as "ad title", "delivery description", etc.) reported by different platforms or channels during data synchronization; the character matching degree refers to the text consistency ratio between the contents of the same field in different channels, which is often measured by character-level comparison (such as Levenshtein distance, similarity percentage); the field that does not meet the standard refers to a field whose content consistency does not meet the pre-set standard, and such fields are considered as high-risk objects of data conflicts or abnormalities; the pending review mark refers to the "pending manual review" or "further processing" status mark automatically applied to the field detected by the system as having data conflicts or abnormalities.

[0023] In module 5, the most optimal section of the reliable field refers to the most reliable data content among all the fields reported by different channels, which is obtained through multi-factor comprehensive analysis such as modification frequency and generation time; the similarity distribution refers to the statistical distribution of the text similarity between the contents of the fields of different channels and the reliable field, which is used to assist in determining whether they can be automatically merged; the synchronization standard refers to the consistency and completeness requirements that need to be met for data merging and synchronization of different channels, such as the similarity needs to be higher than a certain set percentage.

[0024] Please refer to Figure 2 , the resource timing scheduling module includes: The resource overlap analysis submodule analyzes resource attributes and user distribution based on the advertisement delivery plan form, judges the intersection of the region and interest label of the user group of each delivery channel, compares the overlap degree of the resource delivery time points, calculates the repeated contact frequency of the intersection user group in the difference platform, and obtains the resource cross-coverage proportion; When receiving the advertisement delivery plan form, for each advertisement resource record, the display position, advertisement type, distribution budget and channel of the resource itself are extracted in turn, and the user target data in the form is further classified to obtain the number of users in different regions and interest labels for each channel user. Assuming that there are 3500 Beijing users and 1800 sports interest users in A channel, and 2800 Beijing users and 1200 sports interest users in B channel, then the numbers of Beijing sports interest users in A and B channels are compared with each other, and the same batch of user numbers existing in both channels are screened out to obtain 800 intersection users. Then, the scheduling time period of all advertisement resources is recorded uniformly, for example, A resource 18:00-19:30 and B resource 19:00-20:00. The start and end times of all resources are divided into 5-minute time slices, and the overlapping interval is marked between 19:00-19:30. In this time period, the intersection users receive the delivery of A and B advertisements respectively. Each user is checked whether there is repeated delivery record in this interval. If there is, the repeated contact number of this user in this time period is accumulated by 1. Assuming that the total repeated contact of the 800 intersection users in this overlapping section is 2200 times, the total repeated contact number of each user in all overlapping sections is counted, and finally, the total repeated contact number of all intersection users is directly compared with the number of intersection users. If 2200 contacts correspond to 800 people, then each intersection user is contacted 2.75 times on average in all overlapping sections. Through this statistical method, the resource cross-coverage proportion is obtained.

[0025] The traffic interval comparison submodule analyzes the matching of traffic nodes and audience labels based on the resource cross-coverage proportion, compares the overlap of user attention labels and resource traffic in the difference period, and counts the distribution of high-frequency labels in the active traffic interval to obtain a label overlap density factor. The traffic monitoring records of each resource in the actual advertising data are collected time period by time. For the traffic value of each ad position every 5 minutes, the distribution of all delivery resources in high-traffic and low-traffic periods is first sorted out, and the resource traffic distribution related to the intersection users is found. If the total traffic of resource A during 19:00-19:30 is 24,000, and that of resource B is 15,000, the repeated exposure traffic of the intersection users in the two resources in this time period is filtered out, and the user distribution of each traffic data and the corresponding interest tag is further analyzed. The interest tags of the intersection users, such as sports, beauty, digital, etc., are matched one by one. Go to each traffic node and count the actual number of users of each interest tag in different time periods. For example, from 19:00 to 19:30, there are 1,100 users interested in sports and 300 users interested in beauty. Active users above 1,000 are defined as high-frequency tags, and sports interest tags are recorded as high-frequency. All high-traffic nodes are traversed, and the number of times the sports interest tag appears in high-traffic nodes is accumulated. Assuming that the sports interest tag appears 7 times in 10 high-traffic nodes, directly divide 7 by 10 to calculate the tag overlap density factor of 0.7. Finally, the tag overlap density factor is obtained in this way.

[0026] The time sequence adjustment submodule analyzes the marginal gain changes and display frequency of each resource based on the label overlap density factor, compares the distribution differences between features and time series, and uses the formula: ; Get the time structure change , adjust the time node sequence of resources to obtain the resource delivery overlap characteristics, where, Indicates the The cross-coverage ratio of resources in a region reflects the degree of intersection between user groups in different regions. Indicates the The average tag frequency of the user group in a region is used to reflect the appearance density of users' active tags. Indicates the The time span of resource delivery in a region, that is, the duration of resource delivery in the region. represents the number of regions involved in the analysis, Indicates the The marginal gain change index of each resource in the overlapping interval reflects the gain change trend brought about by resource input. Indicates the The display frequency of a resource per unit time is used to describe the density of resource delivery. Indicates the number of resources participating in the analysis.

[0027] The time structure variation is a statistical index for quantifying the arrangement difference, superposition strength and optimization reconstruction degree of all advertising resources in the time dimension. The greater the index value, the more obvious the variation of the overall delivery time structure, the resource time intersection and misplacement. The smaller the index value, the smaller the difference between the optimized reconstruction resource delivery time sequence and the original time sequence, and the weaker the overlapping or misplacement degree. It is a quantitative result for measuring the optimization reconstruction amplitude of the advertising resource scheduling in the time axis.

[0028] The resource Z1 to Z4 is collected under the condition of multiple platforms, and the data of the new click number change curve obtained under the unit budget increase is calculated, and the marginal gain variation index is calculated , for example, the original marginal gain of Z1 is 2.1, and the normalized value is 1.05, the marginal gain of Z2 is 1.7, and the normalized value is 0.85, the marginal gain of Z3 is 2.3, and the normalized value is 1.15, the marginal gain of Z4 is 1.9, and the normalized value is 0.95; the corresponding unit time display frequency is calculated by recording the total number of displays and the total duration of the platform, wherein the original display frequency of Z1 is 0.7, and the normalized value is 0.875, the display frequency of Z2 is 0.9, and the normalized value is 1.125, the display frequency of Z3 is 0.6, and the normalized value is 0.75, the display frequency of Z4 is 0.8, and the normalized value is 1.0; at the same time, the cross coverage of the advertising resources in the regions R1 to R3 and the label response frequency of the users are counted respectively, and the data of and are obtained, wherein the of R1 is normalized to , , the normalization of is , the normalization of is , the normalization of is , the normalization of is , the normalization of is , the normalization of is , the normalization of is , the normalization of is According to the above data, the formula is calculated: ; ; ; The left sum is: ; Right side part: ; ; ; ; Right side sum result: ; Final operation result: ; The results show that the time allocation structure of the current resources in multiple platforms and regions has obvious deviation, and the display frequency and marginal gain of some resources are concentrated in the high coincidence area, which fails to achieve coordinated optimization with the user label active period. Reflects the structural variation amplitude between each allocation window in the time series of resources, the larger the value, the more obvious the aggregation and misplacement of resources in the time dimension, which needs to be adjusted by sorting to reduce resource allocation redundancy. The formula compares the time scheduling constructed by and two parts, not only integrates user label response and resource coverage breadth, but also includes resource response strength and consumption speed, so as to more accurately reflect the coincidence degree of time series structure and quantify the necessity of optimization adjustment.

[0029] Please refer to Figure 3 , the keyword screening module includes: The portrait active extraction sub-module identifies the user portrait information under the product label based on the resource allocation coincidence feature, analyzes the access frequency of user groups in each time period, judges the access time period distribution of each type of user, determines the time period of access peak and active set, and obtains the proportion of user active time period. Read all the delivery resources in different time period of coincidence attribute, for each product label corresponding to the advertising resources, filtering the user number and its user portrait data associated in the delivery plan, divide the data into age, region, interest preference, gender and other attribute labels, and aggregate according to the product label, then divide the user set under each product label into access frequency per hour within 24 hours, count the login, click, browse and other behavior times of users per hour, for example, for the users under the sports shoes product label, count 200 users accessing from 0:00 to 1:00, 2400 users accessing from 18:00 to 19:00, accumulate the access data of each hour, and summarize the access behaviors of all users in different time periods, compare the proportion of the number of users accessing in each time period to the total number of users, analyze the distribution of each user group in each hour, and calculate the proportion of the number of users accessing in the 18:00-22:00 time period to the total number of users for the 18-24 age group, assuming that the number of users accessing in this time period is 1200 and the total number of users in a day is 3000, the proportion of this time period is 40%, further determine which time period the proportion of the number of users accessing is higher than the set active threshold, if the active threshold is set to 30% of the total number of users, then 40% belongs to the peak section, mark the time period higher than the threshold as the user active time period, iterate all product labels and all user portraits according to this method, record all time periods in the peak and active set as the user active time period, finally, calculate the proportion of the number of users in the section to the total number of users to get the proportion of user active time period under each product label.

[0030] The keyword coincidence comparison sub-module compares the coincidence of the active time period of the keyword and the time node of the advertisement delivery according to the proportion of the user active time period, judges the coverage range of the coincident time period, selects the keywords with high time coincidence degree, and gets the time coincidence range of the keywords. First, the total click behavior record of the keyword is called, the click time period of each keyword within 24 hours is segmented, the number of clicks of the keyword in each time period is accumulated, and the click heat distribution is formed by hour statistics. Then, the active time period of each keyword is compared with the actual time node of the advertisement placement schedule, the active time period of each keyword is aligned with the time period of the advertisement schedule, the interval overlap determination method is used, and whether there is an intersection between the active time period of the keyword and the actual placement time period of the advertisement is judged hour by hour. If a keyword is active from 18:00 to 21:00 and the advertisement placement schedule is from 17:00 to 20:00, the overlapping time period is from 18:00 to 20:00. The number of clicks of all keywords in the overlapping period is counted, and the proportion of the number of clicks in the overlapping period to the total number of clicks of the keyword during the day is calculated. For example, the keyword "sports shoes" clicks 2000 times from 18:00 to 20:00, and the total number of clicks during the day is 5000, the overlapping click proportion is 40%, and the screening keyword time overlap threshold is set to 30%. The keywords with a higher overlap ratio than the threshold are selected as high overlap keywords, and the keywords with a high time overlap degree are sorted by proportion from high to low. The time overlap degree is calculated, and the keywords in the top list and their overlapping time period are output as the keyword time overlap degree.

[0031] The click frequency sorting submodule calls the keyword time overlap degree, optimizes the budget configuration and click performance of the keyword, compares the click efficiency and user coverage of the unit budget, calculates the sorting factor of each keyword, and uses the formula: ; The keywords are arranged in order according to the sorting order to obtain a keyword sorting label set, wherein, represents the sorting factor of the th keyword in the sorting label set, represents the click frequency of the th keyword in the overlapping period with the advertisement placement time, represents the budget configuration of the th keyword, represents the click frequency of the th keyword and the th user group in the overlapping period, represents the budget influence degree of the th user group, represents the click active offset of the th keyword in the overlapping period, represents the budget consumption of the th keyword in the non-overlapping period, represents the total budget configuration of the th keyword, represents the number of user groups.

[0032] The sorting factor is a composite reference index for comprehensively evaluating and sorting the priority of each keyword bid. By integrating the click performance of the keyword in the overlapping period of the advertisement bid, the budget configuration, the interaction frequency with different user groups, the budget impact, the click active offset, and the budget consumption, etc., the overall priority and resource utilization efficiency of each keyword in the current advertisement bid environment are reflected, so as to determine the order of the keywords in the bid list.

[0033] For example, the budget configuration data of the keyword "running shoes" in the bid platform is extracted. The total budget configuration of the keyword is 500 yuan, and the non-overlapping time period budget consumption is 110 yuan, which is normalized as budget configuration , non-overlapping budget , and the click frequency of the keyword in the overlapping period covered by the bid plan is 280 times, which is normalized as , then the click active offset degree of the keyword in the user group is analyzed, the click concentration offset is calculated as 0.7 according to the click fluctuation interval, which is normalized as , and then the click frequency of the keyword in each user group is split, which is 55 times for user group 1, 63 times for user group 2, and 41 times for user group 3, which is normalized as , , , the corresponding group budget impact degree is: , , ; After normalization, the proportion remains unchanged and is substituted into the formula , that is: ; The square root of the keyword budget configuration is calculated: ; Substitute the above normalized results into the formula: ; ; The result shows that the calculated The numerical value of the keyword "running shoes" in the advertisement placement scheme based on the coincidence time period click frequency, budget configuration, user group interaction and budget consumption, etc. The larger the sorting factor value, the better the advertisement placement efficiency and resource utilization efficiency of the keyword in the target audience active period, so in the subsequent steps, all candidate keywords can be ranked in descending order, and finally the keyword sorting tag set is obtained, thereby providing data support for precise placement decision of the advertisement system. After normalization processing of multiple difference dimensions, the formula avoids the imbalance between budget amount and behavior click frequency affecting the accuracy of sorting, and introduces a non-coincidence period loss term to strip inefficient resource waste, so that the sorting index can truly reflect the comprehensive matching adaptability and resource allocation rationality of the keyword.

[0034] Please refer to Figure 4 The placement path construction module includes: The audience attribute analysis submodule identifies the associated audience portrait features and product attribute content based on the keyword sorting tag set, analyzes the correspondence of gender, age and interest audience tags with product categories, compares the population distribution under different tags with keyword combinations, and obtains the tag population distribution information. Each tag in the keyword tag set is disassembled to extract the corresponding keyword, preferred level and active period information of the tag. Then, for each keyword tag, the portrait information of the associated users is retrieved one by one, the gender, age and interest tag parameters of the users are called, and the product browsing or purchase attributes corresponding to each user are synchronously captured. For example, under the "sports shoes" tag, 3200 male users and 1800 female users are read, 2200 users in the 18-24 age group and 1500 users in the 25-34 age group are obtained respectively, the association search of the keyword and the interest tag is performed on the interest attribute, and the number of users associated with "sports shoes" and "sports" and "outdoor fitness" interest tags is counted. In the product category dimension, the number of users and keyword click data under each gender, age group and interest combination are paired and compared. The number of users under each group and the keyword click volume, browsing volume and conversion behavior are listed in groups by direct comparison. Assuming that the number of users under the combination of 25-34 years old, female and outdoor fitness interest is 900, and the keyword "anti-slip sports shoes" is clicked 320 times, the product category tag is "running shoes", and the comparison result shows the corresponding relationship between the product category and the interest, gender and age tags under this combination. The user distribution under each tag group is classified and counted, and the tag group higher than the average of the overall tag distribution is set as the main force user distribution, and the tag group lower than 30% of the overall average is set as the weakly related tag group. The distribution under all difference tags and keyword combinations are compared one by one, and the main force user tag group and its matching keyword are screened out. All tag combinations and user number distribution are summarized to obtain the tag population distribution information.

[0035] The resource matching screening sub-module analyzes the matching degree of the product label and the keyword combination of each resource node according to the label population distribution information, judges the time overlap of the keyword active period and the resource delivery window, screens the resource nodes with consistent labels and sufficient time intersection, and obtains the node label coincidence list; The product label corresponding to each advertising resource node and the deliverable keyword combination are extracted, all nodes are sorted according to their allocated product labels, keyword combinations and audience portrait labels, then the advertising delivery scheduling window of each node is read, the active period of the keyword is compared with the delivery window time period of the resource node in turn, the time interval overlap determination method is adopted, if the active time period of the keyword "smart sports shoes" is 19:00-23:00 and the deliverable time of the resource node is 20:00-22:00, the overlapping interval is 20:00-22:00, all keywords and resource nodes in the overlapping interval are recorded, the matching degree is judged, the label consistency and time overlap length are used as double determination standards, the time overlap sufficient threshold is set as 60% of the actual overlap length to the total length of the delivery window, if the actual overlap is 2 hours and the total delivery window is 3 hours, the overlap ratio is 66%, which exceeds the threshold, the node is marked as time overlap sufficient, all nodes and keyword combinations are traversed, all nodes with completely consistent labels and time overlap ratio higher than the threshold are screened, the node and label, time overlap relationship of all screened nodes are counted, and the node label coincidence list is obtained.

[0036] The user coverage statistical sub-module obtains the target user group information according to the node label coincidence list, analyzes the user behavior data in the keyword active period, classifies and compares the user group quantity under each channel, time period and region, and obtains the path user coverage distribution. Call all the target user number set corresponding to the filtered node, pull all the user numbers associated with each node to the current statistical set, and then classify and count the behavior data of each target user in the keyword active period, group by hour interval, by channel, and by region, record the number of user behaviors and active users in each group, for example, the active users of the "anti-slip sports shoes" keyword in A channel from 19:00 to 21:00 are 780, the active users in B channel during the same period are 450, the active users in Beijing are 320, and the active users in Shanghai are 210. The user number in different regions and channels is counted and then all the data is summarized to the label dimension. The number of users in the same label under different time periods, different channels, and different regions is compared, and the channel or region group with higher activity than the average value of the node is distinguished. Assuming that the average number of users covered by the node is 500, if the number of users in Beijing is 320, it is lower than the average value, and the number of users in Shanghai is 210, which is also lower than the average value. If the number of users in A channel is 780, it is higher than the average value. Mark all the channels or region groups that are higher than the average value. Finally, the number of user groups under each label combination, each time period, each channel, and each region is counted to obtain the path user coverage distribution.

[0037] Please refer to Figure 5 , the data conflict comparison module comprises: The text matching comparison submodule analyzes the text content of each channel synchronization field based on the path user coverage distribution, compares the similarity of character content from different sources under the same field category, judges the matching difference of each group of field content, calculates the similarity interval of each group of fields, and obtains the field matching deviation degree; Call all the channel-pushed synchronization field content, extract all the text content under the same field category in each channel, such as ad title, ad description, etc. Compare the characters of the ad titles of A channel and B channel one by one, check each character and position, count the number of completely consistent characters and the total number of characters, record the number of different characters and the difference position for each group of fields, and compare the matching of "cool and breathable anti-slip sports shoes" in A channel and "cool and breathable anti-slip sports shoes" in B channel. It is found that "cool" and "cool" differ by 2 characters, the total character length is 12, and the difference rate is recorded as 2 / 12. Then, the contents of all field categories are compared two by two according to the source channel, the field contents from different sources are compared, the number of different characters and the total length of each group of fields are accumulated to obtain the similarity percentage of each group of fields. For each comparison result, judge the similarity interval, set the interval standard as follows: more than 90% for high similarity, 70%-90% for medium similarity, and less than 70% for low similarity. The similarity interval of each group of fields is directly classified, the number of fields in different similarity intervals is counted, and finally the field difference rate below the interval threshold in the comparison is recorded and the matching deviation degree of all fields is counted.

[0038] The conflict field screening submodule screens the fields that do not meet the unified comparison standard based on the field matching deviation degree, judges the belonging source, field category and content of the fields, analyzes the field distribution and state corresponding to the screening result, sets the screening field as a to-be-reviewed mark, and obtains a to-be-reviewed field set; All fields lower than the unified comparison standard are screened, and fields with a similarity lower than 70% are regarded as preliminary abnormal objects. The source channel, field category and actual text content of the fields are judged one by one. The content of the A-channel advertisement description "new summer men's anti-slip sports shoes, cool and comfortable" and the B-channel "summer anti-slip sports shoes, breathable and comfortable" is compared, and the similarity is only 55%. The field is included in the abnormal range, and the source channel of each abnormal field is further analyzed. The distribution of the advertisement title, advertisement description, display language and other categories is recorded respectively. The distribution number of the abnormal fields under each channel and field category is classified and counted, and the content difference between the text content of each abnormal field and the standard template is judged. If the content of the abnormal field appears the phenomenon of keyword omission, description theme change and the like, the field is further marked as to-be-reviewed, and the number, source, field category and state of all fields judged to be to-be-reviewed are archived to form a to-be-reviewed field set.

[0039] The abnormal index acquisition submodule analyzes the source, generation time and modification frequency of the fields according to the to-be-reviewed field set, compares the source number under the current review state, and uses the formula: ; The field conflict offset is obtained, the abnormal characteristics of the difference type field are screened, and the abnormal field conflict index is obtained, wherein, represents the field conflict offset of the first group of fields, represents the character content similarity of the first group of fields, represents the generation time sequence number of the first group of fields, represents the modification frequency of the first group of fields, represents the source number of the first group of fields, represents the total number of sources involved in the first group of fields, represents the edit distance of the first group of fields in the first source, represents the field length of the first group of fields in the first source; According to the to-be-reviewed field set, the number of sources, generation time and modification frequency of each field in each channel are analyzed, the character content similarity and edit distance of each field under each source are extracted, each group of fields is numbered as , and the source number is . For field number F102, the number of sources is 3, the generation time is June 15, 2025, the modification frequency is 3 times, and before performing the analysis, each participating item is converted into a normalized dimensionless form, wherein the normalized generation time is set to , the normalized modification frequency is set to , the character similarity is standardized to , the field appears in 3 channels, and is set to , the character edit distances of the three sources are 2.1, 1.5 and 1.9 respectively, and the field lengths are 14, 13 and 15 respectively, which are normalized to , , , the field length is set to , , , after normalizing the parameters, the following formula is used to calculate the field conflict offset: ; ; ; ; The field conflict offset of this field group is , which represents the text difference intensity of the field content under the comprehensive action of source distribution, modification activity, generation time sequence and other dimensions. If the lower limit of the offset recognition interval set according to the field category is 0.12, this item is greater than the judgment boundary and should be classified as an abnormal field, which enters the subsequent processing sequence, records its field number, conflict intensity level, source coverage range and variation frequency, to form a structured output, obtains the abnormal field conflict index, and the formula effectively reveals the non-consistent distribution characteristics in the structure field by normalizing the content similarity ( ), behavior time characteristics ( , ) and platform distribution characteristics ( ), and combining the source edit difference ( ), so that the conflict judgment is no longer dependent on a single dimension, thereby strengthening the integrity and universality of system abnormality recognition.

[0040] Please refer to Figure 6 , the content archiving and auditing module includes: The trusted field identification submodule analyzes the source channel of the field to be reviewed based on the abnormal field conflict indicator, compares the generation time of each field with the historical modification record, judges whether the field meets the requirements of coming from a specified channel and having a concentrated generation time and modification frequency distribution, and obtains a trusted field screening set; All fields marked as to be reviewed are called, the source channel, field content, generation time, and historical modification times of each field are extracted, and a grouped table is established for all field information. For each group of fields, the source channels are compared in turn, the specified channel is set as the authoritative channel list, all fields from the authoritative channel are marked separately, the generation time of each field is read in turn, sorted by time, and fields with a generation time within 30 minutes are determined as a time-concentrated group. Then, the historical modification times of each field are counted, and the modification frequency interval is divided. Fields with a modification frequency of less than or equal to 2 times are classified as a low-frequency group, fields with a modification frequency of 3 to 6 times are classified as a medium-frequency group, and fields with a modification frequency greater than 6 times are classified as a high-frequency group. Combined with the three parameters of source channel, generation time, and modification frequency, it is determined whether the field meets the following conditions: from the authoritative channel, the generation time is within 30 minutes, and the modification frequency is less than or equal to 2 times. If all conditions are met, the field is added to the trusted field screening set. For example, the A-channel field "anti-slip summer sports shoes" comes from the authoritative channel, the generation time is June 2, 2025, 18:25, and the history has been modified only once, which meets the above three conditions and is included as a trusted field. If the B-channel field has a generation time of June 2, 2025, 19:12 and a modification frequency of 5 times, it is not included in the trusted set, forming a trusted field screening set.

[0041] The similarity merging judgment submodule analyzes the character structure of the trusted field screening set and the content of each channel field, compares the differences between the contents, judges whether the character content meets the consistency requirements of the merging criteria, merges the fields that meet the conditions, and obtains a merged field comparison result. The archiving proportion statistical submodule counts the number of merged fields in the overall field according to the merged field comparison result, analyzes the proportion of fields originating from trusted fields, and judges the coverage range of merged fields and trusted fields, to obtain the archiving content coverage proportion. Call the trusted field content, compare the character structure of each trusted field with other channels of the same category field bit by bit, count the number of consistent characters and total characters between them, judge the actual difference between the contents, set the similarity requirement according to the merging standard, set the consistency threshold to 90%, if the character content consistency rate of the trusted field and other channel fields is higher than 90%, it is determined that it meets the merging requirements, all fields are screened one by one, and consistency judgment is performed, if the character consistency rate of A channel "brand new anti-skid summer sports shoes" and B channel "brand new anti-skid summer sports shoes" is 92%, the two fields are merged and archived, if another group of B channel fields only have an 85% consistency rate with the trusted field, no merging is performed, all fields that meet the merging conditions are batch merged, for each pair of merged fields, record the channel, content, merging state, and output the merged field comparison result.

[0042] The number of all completed merged fields is counted, the number of merged fields is directly compared with the total number of fields to obtain the proportion of merged fields, the number of all merged fields directly derived from the trusted field is counted, and the number is compared with the total number of merged fields to obtain the proportion of trusted fields, and the threshold of the proportion of trusted fields is set to 80%, if the total number of merged fields is 120 and the number of trusted fields is 100, the proportion of trusted fields is 83%, which is higher than the threshold, finally, the range of the field categories covered by all the merged fields is counted and compared with the total number of field categories to judge the distribution of the archived content in all categories, if the merged fields cover 8 of the 10 categories, the archiving range is 80%, all the above values are recorded to obtain the archiving content coverage ratio.

[0043] The above is only a preferred embodiment of the present application, and does not limit the form of the present application, any skilled person in the art can modify or change the above disclosed technical content to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification of the above embodiments without departing from the technical solution of the present application, according to the technical essence of the present application, still belongs to the protection scope of the technical solution of the present application.

Claims

1. Cross-platform advertising and SEO collaborative optimization system, characterized by: The system comprises: The resource timing scheduling module analyzes the relationship between resource parameters, user coverage, and time nodes based on the advertising delivery plan form, compares the time overlap intervals of each resource, analyzes the overlap between traffic distribution nodes and user intervals, adjusts the order of resource time nodes, and obtains the resource delivery overlap characteristics; The keyword screening module analyzes product classification tags, audience portraits, and user active time periods based on the resource delivery overlap characteristics, compares the click frequency of entries with the budget, determines the relationship between keyword active time and overlapping areas, selects entries with consistent active time, and sorts them by click frequency to obtain a keyword ranking tag set; The delivery path construction module analyzes the audience portrait and product attributes based on the keyword sorting tag set, optimizes the matching of platform resources with the target group and delivery window in combination with the budget, screens resource nodes with overlapping attributes, counts and classifies the user groups, and obtains the path user coverage distribution; The data conflict comparison module determines the degree of character matching of similar fields based on the path user coverage distribution, filters out fields with substandard matching ratios, assigns them to be reviewed, optimizes abnormal data detection, and obtains abnormal field conflict indicators.

2. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The resource delivery overlap characteristics include time series distribution characteristics, audience overlap characteristics, and delivery structure characteristics; the keyword sorting tag set includes tag identification, priority level, and active period; the path user coverage distribution includes channel distribution characteristics, user quantity distribution, and delivery period characteristics; the abnormal field conflict indicator includes abnormal identification, conflict type, and processing mark.

3. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The resource timing scheduling module includes: The resource overlap analysis submodule analyzes resource attributes and user distribution based on the advertising plan form, determines the intersection of regions and interest tags of user groups in each delivery channel, compares the degree of overlap of resource delivery time points, calculates the frequency of repeated contact of the intersection population on different platforms, and obtains the resource cross-coverage ratio; The traffic interval comparison submodule analyzes the matching of traffic nodes and audience tags based on the resource cross-coverage ratio, compares the overlap between user-focused tags and resource traffic in different time periods, and calculates the distribution of high-frequency tags in active traffic intervals to obtain the tag overlap density factor; The time sequence adjustment submodule analyzes the marginal gain changes and display frequency of each resource based on the label overlap density factor, compares the distribution differences between the features and the time series, obtains the time structure changes, adjusts the time node sequence of the resources, and obtains the resource delivery overlap characteristics.

4. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The keyword screening module includes: The portrait activity extraction submodule identifies user portrait information under the product tag based on the resource delivery overlap characteristics, analyzes the access frequency of user groups in different time periods, determines the access time distribution of each type of user, determines the time periods with peak access and concentrated activity, and obtains the proportion of user active time periods; The entry overlap comparison submodule filters the click behavior of keywords in different time periods based on the proportion of user active time periods, compares the overlap between the keyword active time periods and the advertising delivery time nodes, determines the coverage of the overlapping time periods, and filters keywords with high time overlap to obtain the keyword time overlap breadth; The click frequency sorting submodule calls the time overlap breadth of the keywords, optimizes the budget configuration and click performance corresponding to the keywords, compares the click efficiency and user coverage of the unit budget, calculates the ranking factor of each keyword, arranges the keywords in sorting order, and obtains a keyword sorting tag set.

5. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The delivery path construction module includes: The audience attribute analysis submodule identifies the associated audience portrait characteristics and product attribute content based on the keyword sorting tag set, analyzes the correspondence between gender, age and interest audience tags and product categories, compares the population distribution under different tags with keyword combinations, and obtains tag population distribution information; The resource matching and screening submodule analyzes the matching degree between the product label and keyword combination of each resource node based on the label population distribution information, determines the time overlap between the keyword active period and the resource delivery window, and screens resource nodes with consistent labels and sufficient time overlap to obtain a node label overlap list; The user coverage statistics submodule obtains target user group information based on the node label overlap list, analyzes user behavior data during the keyword active period, classifies and compares the number of user groups under each channel, time period and region, and obtains the path user coverage distribution.

6. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The data conflict comparison module includes: The text matching and comparison submodule analyzes the text content of the synchronized fields of each channel based on the user coverage distribution of the path, compares the similarity of the character content of the different sources under the same field category, determines the matching difference of each group of field contents, calculates the similarity interval of each group of fields, and obtains the degree of field matching deviation; The conflict field screening submodule screens out fields that do not meet the unified comparison criteria based on the degree of field matching deviation, determines their source, field category, and content, analyzes the field distribution and status corresponding to the screening results, sets the screened fields as pending review marks, and obtains a pending review field set; The abnormal indicator acquisition submodule analyzes the source, generation time and modification frequency of the fields according to the set of fields to be reviewed, compares the number of sources in the current review state, obtains the field conflict offset, screens the abnormal characteristics of the difference type fields, and obtains the abnormal field conflict indicator.

7. The cross-platform advertising and SEO collaborative optimization system according to claim 1 is characterized in that: The system further comprises: The content archiving and review module analyzes the source channel, generation time, and modification frequency of the fields to be reviewed based on the abnormal field conflict indicators, identifies credible fields as standards, compares the similarity of field content in each channel, merges fields that meet the standards, and calculates the proportion of merged archived fields to obtain the archived content coverage ratio; The archived content coverage ratio includes the merge ratio, the trustworthy ratio, and the archive scope.

8. The cross-platform advertising and SEO collaborative optimization system according to claim 7, characterized in that: The content archiving and review module includes: The trusted field identification submodule analyzes the source channels of the fields to be reviewed based on the abnormal field conflict indicators, compares the generation time of each field with the historical modification records, determines whether the fields meet the requirements of being from the specified channel, having a concentrated generation time, and a modification frequency distribution, and obtains a trusted field screening set; The similarity merging judgment submodule screens the trusted fields based on the set, analyzes the character structure of the trusted fields and the content of the fields in each channel, compares the differences between the contents, determines whether the character content meets the consistency requirements of the merging benchmark, merges the fields that meet the conditions, and obtains the merged field comparison results; The archiving ratio statistics submodule counts the number of merged fields in the overall field based on the merged field comparison result, analyzes the proportion of them derived from credible fields, and determines the coverage of merged fields and credible fields to obtain the archiving content coverage ratio.

Images