A multi-platform adapted AI marketing content intelligent delivery control system

Abstract

The application discloses a multi-platform adapted AI marketing content intelligent putting control system, and relates to the technical field of platform marketing putting, solves the problem that data transmission is lagged and putting efficiency and user experience are affected due to code rate mismatch, and lays a foundation for efficient putting by accurately adapting to the data transmission requirements of each platform and avoiding the problems of poor data transmission and resource waste caused by code rate mismatch through deep analysis of code rates of different platforms and different data formats, locking of characteristic code rate intervals, and innovative construction of an optimal sorting set according to the characteristic code rate intervals and data capacity, and optimization of transmission order through variance processing and other methods, which greatly improves data transmission efficiency and guarantees that marketing content can reach target users at the fastest speed, and enhances putting timeliness.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of platform marketing delivery, in particular to an AI marketing content intelligent delivery control system adapted to multiple platforms. BACKGROUND

[0002] In the current digital marketing boom, enterprises often need to deliver AI marketing content on multiple platforms such as Douyin, WeChat, Baidu, and Taobao to expand their market influence.

[0003] However, the existing technology has many problems in the process of multi-platform marketing content delivery.

[0004] On the one hand, different platforms have significantly different requirements for the bit rate of text, video, audio, and other data formats. Traditional delivery systems lack precise analysis of bit rate characteristics, often resulting in data transmission lag due to mismatched bit rates, affecting delivery efficiency and user experience, and wasting network resources.

[0005] On the other hand, the transmission sequence of marketing content is not reasonable, and the relationship between data capacity and transmission time is not fully considered, making it impossible to achieve fast data transmission, resulting in marketing information not reaching users in a timely manner.

[0006] In addition, the active periods of users on different platforms are different, but existing delivery technologies cannot accurately grasp the optimal push period, resulting in low attention and poor conversion rate, and low utilization of marketing resources.

[0007] Therefore, there is an urgent need for an AI marketing content intelligent delivery control system that can adapt to the characteristics of multiple platforms, optimize data transmission, and accurately determine the push period to improve marketing effectiveness and business value. SUMMARY

[0008] To address the shortcomings of existing technology, the present application provides an AI marketing content intelligent delivery control system adapted to multiple platforms, which solves the problem of data transmission lag caused by mismatched bit rates, affecting delivery efficiency and user experience.

[0009] To achieve the above purpose, the present application is implemented by the following technical solution: an AI marketing content intelligent delivery control system adapted to multiple platforms, comprising:

[0010] A historical data processing end acquires platform data adapted to the current AI from cloud data, the acquired platform data including bit rates associated with different data formats, and from a large amount of processed data, the characteristic bit rate interval associated with the corresponding data format is locked, in particular:

[0011] Confirming the platform adapted by the current AI, obtaining the platform data belonging to the platform from the cloud data, and confirming the average code rate M of the corresponding data format in the single data transmission process from the platform data i , where i represents different data formats, and the average code rate M of several groups for a single data format in the past 30 days i Performing clustering processing:

[0012] Randomly selecting a group M i from several average code rates M i as the intermediate value, and then locking other M i and the check difference value associated with the intermediate value, and if the check difference value is less than or equal to Y1, the corresponding M i is recorded as the clustering value associated with the corresponding intermediate value, where Y1 is a preset value.

[0013] Recording the total number of clustering values G k associated with different intermediate values, where k represents different intermediate values, and selecting the intermediate value of G k max from the recorded total number of clustering values G k is recorded as the selected feature, and the minimum and maximum values of the several clustering values of the selected feature are confirmed, the feature code rate interval for the current platform for the current data format is locked, and the feature code rate intervals for different data formats are confirmed in turn.

[0014] Confirming the feature code rate intervals associated with different data formats for other platforms in turn.

[0015] The feature content sorting end confirms different data formats associated with the current marketing content and classifies the content, and sorts the current classified content, generates the optimal sorting set belonging to the current platform, and the specific method is:

[0016] Confirming that the current marketing content belongs to different classified content of different data formats, and marking the data capacity associated with different classified content as R i , and locking the time interval associated with the corresponding data format according to the feature code rate interval associated with the corresponding data format.

[0017] Based on the time interval associated with different classified content, it is confirmed whether there is an integer time in the corresponding time interval, if there is, the current classified content is recorded as the pending content, if there is not, the corresponding classified content is recorded as the pending combination content, the interval intermediate value of the corresponding time interval of different pending combination content is confirmed, and the confirmed interval intermediate value is sorted in ascending order according to the numerical value, and the several groups of pending combination content are sorted to confirm the content set.

[0018] And from the confirmed content set, several groups of to-be-combined contents are combined two by two, and in the combination process, the two groups of time intervals associated with the two groups of to-be-combined contents are combined and verified, and a to-be-combined time is randomly selected from the two groups of time intervals, if there is an integer time that is the sum of the two groups of to-be-combined time, then the corresponding combination is recorded as an execution combination, otherwise, no record is made;

[0019] A plurality of different combination processes are performed, the difference between the two interval values corresponding to the execution combination in each combination process is confirmed, the difference is greater than or equal to 0, and the difference values associated with the plurality of execution combinations are processed by variance, the standard deviation of the corresponding combination process is confirmed, and the minimum value is selected from different standard deviations associated with different combination processes. The combination process associated with the minimum value is used as the optimal process;

[0020] The plurality of execution combinations associated with the optimal process are used as selected combinations, and the other classified contents that have not been combined are sorted according to the sorting method of the content set to confirm the previous content set, and then the execution combinations are sequentially divided into subsequent positions of the previous content set to generate an optimal sorting set belonging to the current marketing content, and the generated optimal sorting set is transmitted through the transmission execution center.

[0021] The push period determination end confirms the user quantity change data associated with the corresponding platform in the past period from the cloud data to generate a user quantity change curve, and selects the optimal push period of the corresponding platform from the user quantity change curve, and the specific method is:

[0022] Taking the current time as the calibration time, the user quantity change data associated with the corresponding platform in the past 24 hours is confirmed from the cloud data, and the user quantity change curve associated with the corresponding 24 hours is generated according to the different user quantities associated with different time nodes;

[0023] A single curve segment is randomly selected from the user quantity change curve, and the curve characteristics of the single curve segment are confirmed, the user quantities associated with the curve segment are processed by mean value to confirm the first feature T1, the user quantities associated with the two endpoints of the curve segment are marked as Yq1 and Yh1, Yq1 represents the user quantity associated with the initial endpoint of the corresponding curve segment, Yh1 represents the user quantity associated with the terminal endpoint of the corresponding curve segment, the user quantity associated with the point before the initial endpoint is marked as Yq2, and the user quantity associated with the point after the terminal endpoint is marked as Yh2, and then the two groups of difference values Cz1 and Cz2 are summed to lock the second feature of the curve segment, and the comprehensive feature of the corresponding curve segment is locked by using: comprehensive feature = first feature × C1 + second feature × C2, wherein C1 and C2 are both preset fixed coefficient factors;

[0024] Confirm different comprehensive features associated with different curve segments selected, and select the curve segment associated with the maximum comprehensive feature, and take the time period associated with the curve segment as the optimal push time period of the corresponding platform;

[0025] The transmission execution center controls the corresponding platform to push the associated marketing content within the optimal push time period according to the optimal push time period and the optimal sorting set confirmed by the corresponding platform.

[0026] Preferably, the check difference value = |M i - the intermediate value |.

[0027] Preferably, if the check difference value > Y1, no processing is performed.

[0028] Preferably, after the transmission execution center transmits the optimal sorting set to the corresponding platform, the marketing content is reorganized to obtain marketing content.

[0029] The application provides a multi-platform adaptive AI marketing content intelligent delivery control system. Compared with the prior art, the following beneficial effects are achieved:

[0030] At the data processing level, the historical data processing end can accurately adapt to the data transmission requirements of each platform by deeply analyzing the code rate of different platforms and different data formats, locking the feature code rate interval, avoiding the problem of poor data transmission or resource waste caused by unmatched code rate, and laying a foundation for efficient delivery.

[0031] The feature content sorting end innovatively constructs an optimal sorting set according to the feature code rate interval and the data capacity, optimizes the transmission order through variance processing and other methods, greatly improves the data transmission efficiency, ensures that the marketing content can reach the target user at the fastest speed, and enhances the timeliness of delivery.

[0032] The push time period determination end uses user quantity change curve and comprehensive feature analysis to accurately locate the optimal push time period of each platform, so that the marketing content reaches the target user at the moment when the user activity is the highest, significantly improves the user attention and marketing conversion rate, effectively improves the utilization rate of marketing resources, helps enterprises to realize accurate and efficient delivery in multi-platform marketing, and obtains better marketing effect and business value. BRIEF DESCRIPTION OF DRAWINGS

[0033] Figure 1 The figure is a schematic diagram of the principle framework of the application;

[0034] Figure 2 The figure is a determination method of the clustering value of the application. DETAILED DESCRIPTION

[0035] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present application.

[0036] First embodiment

[0037] Please refer to Figure 1 The application provides an AI marketing content intelligent delivery control system adapted to multiple platforms, comprising a historical data processing end, a feature content sorting end, a transmission execution center and a push time period determination end, wherein the historical data processing end, the feature content sorting end and the transmission execution center are electrically connected in sequence from the output node to the input node, and the push time period determination end is electrically connected with the input node of the transmission execution center.

[0038] The historical data processing end obtains platform data adapted to the current AI from cloud data, the obtained platform data comprises a code rate associated with different data formats, locks a feature code rate interval associated with the corresponding data format from a large amount of processed data, and transmits it to the feature content sorting end. The so-called feature code rate interval is a numerical interval with the most clustered corresponding code rate. After determining such a numerical interval, the related feature content can be effectively sorted for subsequent transmission to the current platform.

[0039] The specific way of locking the feature code rate interval is as follows:

[0040] Confirm the platform adapted to the current AI, obtain the platform data belonging to the platform from the cloud data, and confirm the average code rate M i of the corresponding data format in the single data transmission process from the platform data, wherein i represents different data formats. i Cluster processing:

[0041] In combination with Figure 2 , a group of M i is randomly selected from the several average code rates M i as the intermediate value, and the other M i is locked in turn, and the check difference value associated with the intermediate value is greater than or equal to 0 (check difference value = |M i -intermediate value |), if the check difference value is less than or equal to Y1, the corresponding M i is recorded as the clustering value associated with the corresponding intermediate value, wherein Y1 is a preset value, and its specific value is determined by the operator according to experience, otherwise, no processing is performed.

[0042] the total number of cluster values associated with different intermediate values k recorded, where k represents different intermediate values, and the total number of cluster values G recorded k G is selected k The intermediate value of max is recorded as the selected feature, and the minimum and maximum values of the selected feature are confirmed, the feature code rate interval belonging to the current platform for the current data format is locked, and the feature code rate intervals associated with different data formats of other platforms are confirmed in turn;

[0043] In succession, the feature code rate intervals associated with different data formats of other platforms are confirmed. Specifically, the current AI has multiple adaptation platforms, including Douyin, WeChat, Baidu, and Taobao, etc. Different platforms have different code rate features for different data formats, such as text format, video format, and audio format, etc. Therefore, the code rate features associated with the corresponding data format are confirmed from the specific historical data, so as to facilitate subsequent feature sorting of the current push content.

[0044] Among them, the feature content sorting end confirms the different data formats associated with the current marketing content and classifies the content, and sorts the features of the current classified content to generate the optimal sorting set belonging to the current platform. Specifically, the optimal sorting set is that when the actual transmission process is performed according to the sorting set, the specific rate of the data transmission process can be guaranteed, so as to achieve the optimal transmission processing effect.

[0045] The specific generation method of the optimal sorting set is:

[0046] Confirming that the current marketing content belongs to different classified content of different data formats, and marking the data capacity associated with different classified content as R i According to the feature code rate interval associated with the corresponding data format, the time interval associated with the corresponding data format is locked. Specifically, the endpoint value of the time interval is the corresponding data capacity divided by the endpoint value of the corresponding code rate interval. According to the specific value obtained, the time interval belonging to the corresponding platform can be generated.

[0047] Based on the time interval associated with different classified content, it is confirmed whether there is an integer time in the corresponding time interval. If there is, the current classified content is recorded as a pending content. If there is not, the corresponding classified content is recorded as a pending combination content. The interval intermediate value of the corresponding time interval of different pending combination contents is confirmed, and the confirmed interval intermediate value is sorted in ascending order according to the numerical value. The several groups of pending combination contents are sorted to confirm the content set.

[0048] And from the confirmed content set, several groups of to-be-combined contents are combined two by two, and in the combination process, the two groups of time intervals associated with the two groups of to-be-combined contents are combined and verified, and a to-be-combined time is randomly selected from the two groups of time intervals, if the sum of the two groups of to-be-combined time belongs to an integer time, the corresponding combination is recorded as an execution combination, otherwise, no record is made;

[0049] A plurality of different combination processes are performed, the difference between the two interval values corresponding to the execution combination in each combination process is confirmed, the difference is greater than or equal to 0, and the difference values associated with the plurality of execution combinations are processed by variance, the standard deviation of the corresponding combination process is confirmed, and the minimum value is selected from the different standard deviations associated with different combination processes, and the combination process associated with the minimum value is selected as the optimal process;

[0050] The plurality of execution combinations associated with the optimal process are selected as the selected combination, and the other classified contents that have not been combined are sorted according to the sorting method of the content set, the pre-order content set is confirmed, and then the execution combinations are sequentially divided into subsequent positions of the pre-order content set to generate the optimal sorting set belonging to the current marketing content, and the generated optimal sorting set is transmitted through the transmission execution center.

[0051] Specifically, the marketing content is classified according to different data formats, so that it is divided into a plurality of associated classified contents, and the time interval of the classified content has an interval value, which is sorted in ascending order according to the value, and the corresponding content set is confirmed;

[0052] In the combination processing process, the selected method is faster when randomly combining according to the sorting method of the content set, and the process can be quickly in the optimal state by combining from the front end and the rear end of the content set. From the specific classification processing process, the different combination characteristics associated with each different classification process are verified and analyzed (variance processing method), and from the specific processing process of verification and analysis, the optimal sorting set is locked and transmitted, so as to achieve the optimal transmission processing method.

[0053] Second embodiment

[0054] In the specific implementation process of the above embodiment, compared with the above embodiment, this embodiment mainly confirms the push period of different platforms, so that each platform can achieve the optimal push processing effect when pushing content;

[0055] Among them, the push period determination end confirms the user quantity change data associated with the corresponding platform in the past period from the cloud data, to generate a user quantity change curve, and selects the optimal push period of the corresponding platform from the user quantity change curve, and the specific confirmation method of the optimal push period is:

[0056] With the current time as the calibration time, the user quantity change data associated with the corresponding platform in the past 24 hours is confirmed from the cloud data, and a user quantity change curve associated with the corresponding platform in the past 24 hours is generated according to different user quantities associated with different time nodes. The horizontal axis of the curve is the time line, and the vertical axis is the user quantity.

[0057] A single curve segment is randomly selected from the user quantity change curve, and the curve characteristics of the single curve segment are confirmed. The user quantities associated with the curve segment are averaged to confirm the first feature T1. The user quantities associated with the two end points of the curve segment are labeled as Yq1 and Yh1. Yq1 represents the user quantity associated with the initial end point of the corresponding curve segment, and Yh1 represents the user quantity associated with the end end point of the corresponding curve segment. The user quantity associated with the point before the initial end point is recorded as Yq2, and the user quantity associated with the point after the end end point is recorded as Yh2. The difference values Cz1 and Cz2 are calculated using Cz1=Yq1-Yq2 and Cz2=Yh2-Yh1. The second feature belonging to the curve segment is locked by summing the two difference values Cz1 and Cz2. The comprehensive feature is locked using the formula: comprehensive feature=first featurexC1+second featurexC2, where C1 and C2 are both preset fixed coefficient factors, and their specific values are determined by the operator according to experience. C1 generally takes the value 0.684, and C2 generally takes the value 0.316.

[0058] The different comprehensive features associated with the selected different curve segments are confirmed, and the curve segment associated with the maximum comprehensive feature is selected. The time period associated with this curve segment is taken as the optimal push period of the corresponding platform.

[0059] Specifically, in the selection process of the curve segment, different curve segments are associated with different comprehensive features. The maximum value is selected from the different comprehensive features associated with the multiple curve segments, which can fully reflect the most intense user quantity characteristics associated with the corresponding curve segment. Therefore, marketing content can be pushed at the time of the most users, achieving the optimal delivery processing effect.

[0060] Among them, the transmission execution center transmits the optimal sorting set to the corresponding platform according to the optimal push period and the optimal sorting set confirmed by the corresponding platform, and reorganizes the marketing content to obtain the marketing content. After the combination of the marketing content is completed, the corresponding platform is controlled to push the associated marketing content in the optimal push period.

[0061] Some of the data in the above formula are dimensionless numerical calculations, and the contents not described in detail in the specification are all existing technologies known to those skilled in the art.

[0062] The above examples are only used to illustrate the technical method of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present application can be modified or equivalently replaced without departing from the spirit and scope of the technical method of the present application.

Claims

1. A multi-platform compatible AI marketing content intelligent delivery control system, characterized in that, include: On the historical data processing end, the platform data adapted to the current AI is obtained from the cloud data. The obtained platform data includes the bitrate associated with different data formats. From a large amount of processed data, the characteristic bitrate range associated with the corresponding data format is locked. The feature-based content ranking end identifies the different data formats associated with the current marketing content and categorizes the content. It then performs feature-based ranking on the categorized content to generate the optimal ranking set for the current platform. The specific method is as follows: Identify that the current marketing content belongs to different categories of content in different data formats, and label the data volume associated with each category as R. i And based on the characteristic bit rate range associated with the corresponding data format, the time range associated with the corresponding data format is locked; Based on the time intervals associated with different categories of content, confirm whether there are integer times within the corresponding time intervals. If they exist, record the current category content as pending content. If they do not exist, record the corresponding category content as content to be combined. Confirm the midpoint of the time intervals corresponding to different content to be combined, and sort the confirmed midpoints of the intervals in ascending order of value. Sort several groups of content to be combined and confirm the content set. From the confirmed content set, several groups of content to be combined are combined in pairs. During the combination process, the two time intervals associated with the two groups of content to be combined are combined and verified. The time to be combined is randomly selected from the two time intervals. If there are two times that the sum of the two times to be combined is an integer, the corresponding combination is recorded as the combination to be executed. Otherwise, no record is made. Execute several different combined processes, confirm the difference between the intermediate values ​​of the two intervals of the corresponding combined processes in each combined process, and the difference is ≥0. Then, perform variance processing on the differences associated with several combined processes, confirm the standard deviation of the corresponding combined processes, and select the minimum value among the different standard deviations associated with different combined processes. The combined process associated with the minimum value is taken as the optimal process. Several execution combinations associated with the optimal process are selected as the combination, and other uncombined categories are sorted according to the content set sorting method. The preceding content set is confirmed, and the execution combinations are then assigned to the subsequent positions of the preceding content set to generate the optimal sorting set belonging to the current marketing content. The generated optimal sorting set is then transmitted through the transmission execution center. The push time period determination end confirms the user volume change data associated with the corresponding platform in the past period from the cloud data, generates a user volume change curve, and selects the optimal push time period for the corresponding platform from the user volume change curve; The transmission execution center controls the corresponding platform to push the associated marketing content within the optimal push period, based on the optimal push time and optimal sorting set confirmed by the corresponding platform.

2. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 1, characterized in that, The specific method by which the historical data processing terminal locks the feature code rate range is as follows: Identify the platform that the current AI is compatible with, retrieve platform data belonging to that platform from the cloud data, and determine the average bitrate M for the corresponding data format during a single data transmission process from the platform data. i Where i represents different data formats, and M represents the average bitrate M of several sets of data formats over the past 30 days. i Perform clustering: From several average bitrates M i Randomly select a group M i As an intermediate value, other M values ​​are then locked in sequence. i The check difference associated with the median value is ≥0. If the check difference is ≤Y1, the corresponding M will be... i The cluster value associated with the corresponding median value is denoted as Y1, where Y1 is a preset value. The total number of cluster values ​​G associated with different median values k Record the data, where k represents different intermediate values, from the total number G of the recorded cluster values. k In the middle, select G k The middle value of max is recorded as the selected feature, and the minimum and maximum values ​​of several cluster values ​​of the selected feature are confirmed. The feature bitrate range belonging to the current platform for the current data format is locked, and the feature bitrate ranges of different data formats are confirmed in turn. The feature bitrate ranges associated with different data formats on other platforms were confirmed in turn.

3. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 2, characterized in that, The verification difference = |M i -Middle value|.

4. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 2, characterized in that, If the verification difference is greater than Y1, no processing is performed.

5. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 1, characterized in that, The specific method by which the push time period determination terminal confirms the optimal push time period is as follows: Using the current time as the calibration time, the change data of the number of users associated with the corresponding platform in the past 24 hours is confirmed from the cloud data, and the change curve of the number of users associated with the platform in the past 24 hours is generated based on the different number of users associated with different time points. A single curve segment is randomly selected from the user volume change curve, and the curve characteristics of the single curve segment are confirmed. The average value of several user volumes associated with the curve segment is processed to confirm the first feature T1. The user volumes associated with the two endpoints of the curve segment are then labeled as Yq1 and Yh1, where Yq1 represents the user volume associated with the initial endpoint of the corresponding curve segment, and Yh1 represents the user volume associated with the final endpoint of the corresponding curve segment. The user volume associated with the point before the initial endpoint is denoted as Yq2, and the user volume associated with the point after the final endpoint is denoted as Yh2. The following formulas are used: Cz1 = Yq1 - Yq2 and Cz2 = Yh2 - Yh1. The two sets of differences Cz1 and Cz2 are then summed to lock the second feature belonging to this curve segment. The following formula is used: Comprehensive feature = First feature × C1 + Second feature × C2 to lock the comprehensive feature of the corresponding curve segment, where C1 and C2 are preset fixed coefficient factors. The different comprehensive features associated with the selected curve segments are identified, and the curve segment associated with the largest comprehensive feature is selected. The time period associated with this curve segment is taken as the optimal push time period for the corresponding platform.

6. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 1, characterized in that, The transmission execution center transmits the optimal sorting set to the corresponding platform and then reorganizes the marketing content to obtain the marketing content.

7. The AI ​​marketing content intelligent delivery control system with multi-platform adaptability according to claim 1, characterized in that, The historical data processing terminal, the feature content sorting terminal, and the transmission execution center are electrically connected sequentially from the output node to the input node, and the push time period determination terminal is electrically connected to the input node of the transmission execution center.

Images