An artificial intelligence push system based on big data

Abstract

The present invention discloses an artificial intelligence push system based on big data, which relates to the field of information push technology and solves the problem that insufficient analysis of the traffic usage data of the relevant platform will lead to large deviations in the determined matching push timing. The present invention uses two floating values ​​for floating up and down to confirm the traffic usage of different platforms or related web pages in different states, so that the determined traffic change interval is more accurate. Since the standard deviation can show the discreteness of the data, when the number of values ​​corresponding to the floating interval determined based on the floating process are all the closest states, the data discreteness of this situation is the lowest, and the determined related interval is the most accurate. In order to facilitate the relevant confirmation of the subsequent push time period, the accuracy of this interval must be determined first.

Description

Technical Field

[0001] The present invention relates to the technical field of information push, and in particular to an artificial intelligence push system based on big data. Background Art

[0002] Smart push is an application that combines artificial intelligence, knowledge engineering, the Internet, and database technology. By adding intelligent services to smart terminals, user scenario perception is achieved, and useless push information is displayed with reduced authority. By adding intelligent services on the server side, users can judge the matching degree of push information and achieve the best quota allocation under various constraints.

[0003] The patent application with the publication number CN117539881A discloses an artificial intelligence e-commerce push system based on blockchain and big data in the field of e-commerce technology, including: user portrait construction module, store label construction module, distribution storage module, intention collection module, offline push module and online push module; user portrait construction module is used to generate user portrait; store label construction module is used to generate store portrait; distribution storage module is used to store user portrait and store portrait in a decentralized manner through blockchain technology; intention collection module is used to collect user purchase intention; offline push module and online push module are used to determine the push timing. This solution can accurately push e-commerce data according to user intention, and match the push timing according to the user's free time, reducing the disturbance of push information to the user's daily life.

[0004] When pushing relevant information about blockchain and big data, the relevant platforms and APPs generally identify the relevant portraits of the corresponding users based on the relevant usage records of the corresponding users when pushing information, and then select and push information based on this portrait. However, in the actual push process, only the viewing situation of the relevant personnel is considered without fully analyzing the traffic usage data of the relevant platforms, which will lead to a large deviation in the determined matching push timing, which will cause messages to be pushed when the user is using it normally, thereby affecting the relevant users' experience. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides an artificial intelligence push system based on big data, which solves the problem that the traffic usage data of the relevant platform is not fully analyzed, resulting in a large deviation in the determined matching push timing.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an artificial intelligence push system based on big data, comprising:

[0007] The relevant data collection end collects the relevant flow data of different standard periods used by relevant personnel on the relevant platform, and transmits the collected relevant flow data of different standard periods to the relevant data processing end;

[0008] The relevant data processing end processes the collected relevant flow data of each different standard period, generates the flow data fluctuation curve of the corresponding standard period according to the different relevant flow data corresponding to different moments in the same standard period, and then uniformly processes the generated groups of fluctuation curves to determine the high flow change interval, medium flow change interval and low flow change interval of the relevant platform. The specific method is as follows:

[0009] Confirm several groups of related flow data in the same standard period, use the timeline as the horizontal coordinate axis and the flow value as the vertical coordinate axis, and generate the flow data fluctuation curve belonging to this standard period in the coordinate system based on the changes of several groups of related flow data in this standard period;

[0010] Confirm the maximum flow data and the minimum flow data of the flow data fluctuation curve corresponding to the standard period, and randomly select the first floating value and the second floating value from the maximum flow data and the minimum flow data, wherein there is a limiting standard: the second floating value < the first floating value < the maximum flow data, the minimum flow data < the second floating value < the first floating value;

[0011] The first floating value and the second floating value of each group of flow data fluctuation curve are confirmed one by one, so that the first floating value and the second floating value are moved up and down. The moving process cannot exceed the original limit standard, and the first floating value and the second floating value of each flow data fluctuation curve that move up and down are consistent. The number of related flow data between the first floating value and the second floating value in each flow data fluctuation curve at this moment during the moving process is determined and calibrated as G i , and i = 1, 2, ..., n, where i represents different flow data fluctuation curves;

[0012] The several groups G generated at different times i Perform standard deviation processing to determine the standard value BZ, complete a movement process of the first floating value and the second floating value, select the minimum value from the several groups of standard values ​​BZ generated in this movement process, and determine the specific positions of the first floating value and the second floating value corresponding to the minimum value, and determine the medium flow change interval [F1, F2] based on the relevant values ​​F2 and F1 corresponding to the first floating value and the second floating value;

[0013] Select the minimum value Smin from the minimum flow data within the fluctuation curves of several flow data to determine the low flow variation interval [Smin, F1), and then select the minimum value Smax from the maximum flow data within the fluctuation curves of several flow data to determine the low flow variation interval (F2, Smax]);

[0014] The associated time period determination end determines the associated time periods included in the intervals from a number of flow data fluctuation curves based on the determined high flow variation interval, medium flow variation interval, and low flow variation interval, and determines the high time period, low time period, and medium time period based on the intersection of associated time periods in different variation intervals. The specific method is as follows:

[0015] Based on the determined medium flow variation interval, the associated time periods belonging to the variation interval are confirmed from a number of flow data fluctuation curves, and the associated time periods of different flow data fluctuation curves are cross-processed to determine the cross-time periods. Based on the number SG of associated time periods corresponding to the cross-time periods, a group of cross-time periods with the largest number SG is determined as the medium time period. If there is a situation where the number SG is the largest and the same, then both groups of cross-time periods are medium time periods.

[0016] The specific method for determining the low period is:

[0017] Based on the determined low flow variation interval, the low time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period;

[0018] The specific method for determining the high period is:

[0019] Based on the determined high flow variation interval, the high time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period;

[0020] The confirmed high time period, medium time period and low time period are transmitted to the push center;

[0021] The push center determines the number of push messages based on the relevant proportion of the high-time period, and randomly selects from the push database. Combined with the current time, the low-time push unit or the medium-time push unit is used to push the push messages. The specific method is:

[0022] Determine the proportion of high-frequency time periods within 24 hours, and compare the proportion with the preset intervals. There are multiple groups of preset intervals, and the endpoint values ​​of each group of preset intervals are preset in advance. Each group of preset intervals corresponds to a different number of pushes. Based on the preset interval to which the proportion belongs, determine the number of push information TG, and select TG push information from the push database;

[0023] Determine the number of low periods between 0:00 and 24:00:

[0024] If the total number of low-frequency time periods is ≥TG, randomly select push information in the middle of different low-frequency time periods and push the information in sequence until the push is completed;

[0025] If the total number of low time periods is less than TG, a group of push information is selected in the middle time of each subsequent low time period, and when the corresponding low time period arrives, the push information is pushed directly, and the remaining push information is pushed through the subsequent middle time period, and the current time is confirmed, and the time to which the current time belongs is identified:

[0026] If the current time belongs to the low time period, the information will be pushed directly through the low time period push unit, and the time will be monitored. When the current time belongs to the medium time period, the remaining push information will be pushed through the medium time period push unit, and the push time interval will be set in advance, and the push time interval will start from the first information push;

[0027] If the current time belongs to the mid-time period, the remaining push information will be pushed through the mid-time push unit, and the push time interval is set in advance.

[0028] The present invention provides an artificial intelligence push system based on big data. Compared with the prior art, it has the following beneficial effects:

[0029] The present invention uses two floating values ​​to float up and down to confirm the traffic usage of different platforms or related web pages in different states, so that the determined traffic change interval is more accurate. Because the standard deviation can show the discreteness of the data, when the number of values ​​corresponding to the floating interval determined based on the floating process is the closest state, then the data discreteness of this situation is the lowest, and the determined related interval is the most accurate. In order to facilitate the related confirmation of the subsequent push period, the accuracy of this interval must be determined first;

[0030] Based on the different intervals determined, the different time periods associated with the different intervals are determined, and based on the confirmed relevant time periods, information is pushed to ensure that the relevant information pushed can not only achieve a better push effect, but also will not affect the relevant experience of relevant personnel using this platform. The relevant information pushed is more accurate and the effect is better. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 It is a schematic diagram of the principle framework of the present invention;

[0032] Figure 2 It is a schematic diagram of the curve and floating value of the present invention. DETAILED DESCRIPTION

[0033] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0034] See also Figure 1 , the present application provides an artificial intelligence push system based on big data, including a relevant data collection end, a relevant data processing end, an associated time period determination end, a push center and a relevant push database, wherein the relevant data collection end, the relevant data processing end, the associated time period determination end, and the push center are electrically connected in sequence from the output node to the input node, wherein the push database is electrically connected to the input node of the push center;

[0035] Among them, the push center includes a low-time push unit and a medium-time push unit;

[0036] Among them, the relevant data collection end collects the relevant flow data of different standard periods of the relevant personnel using the relevant platform, wherein the standard period is 24h, collects at least five sets of relevant flow data of standard periods, and transmits the collected relevant flow data of different standard periods to the relevant data processing end. Specifically, the relevant platform can be understood as some APPs or some web pages, etc., and its relevant flow data generally has three situations: 1. The situation of large relevant flow represents that the corresponding personnel have relevant behaviors of using or clicking on this platform; 2. The situation of medium relevant flow represents that the corresponding personnel have opened this platform but have not clicked or used it, and there is maintenance flow; 3. The situation of low relevant flow represents that this platform or application is running in the background, and the corresponding personnel have not opened it, but the background running process will also generate relevant flow. Combined with examples: for example, this platform is a game platform, I have opened this game platform and am playing games, then this situation has the largest flow, I have opened this game platform but have not played games, which belongs to the situation of displaying the game platform interface, then it belongs to the state of maintaining flow, the game platform is not opened, and its game platform is running in the background, and this flow state is the lowest;

[0037] The relevant data processing end processes the collected relevant flow data of each different standard period, generates the flow data fluctuation curve of the corresponding standard period according to the different relevant flow data corresponding to different moments in the same standard period, and then uniformly processes the generated groups of fluctuation curves to determine the high flow change interval, medium flow change interval and low flow change interval of the relevant platform. The specific method of uniform processing is as follows:

[0038] Confirm several groups of related flow data in the same standard period, use the timeline as the horizontal coordinate axis and the flow value as the vertical coordinate axis, and generate the flow data fluctuation curve belonging to this standard period in the coordinate system based on the changes of several groups of related flow data in this standard period, such as Figure 2 As shown;

[0039] Confirm the maximum flow data (highest value) and the minimum flow data (lowest value) of the flow data fluctuation curve corresponding to the standard period, and randomly select the first floating value and the second floating value from the maximum flow data and the minimum flow data, wherein there is a limiting standard: the second floating value < the first floating value < the maximum flow data, the minimum flow data < the second floating value < the first floating value;

[0040] The first floating value and the second floating value of each group of flow data fluctuation curves are confirmed one by one, so that the first floating value and the second floating value are moved up and down. The moving process cannot exceed the original limit standard, and the first floating value and the second floating value of each flow data fluctuation curve that move up and down are consistent (that is, when the first floating value in the first group of curves moves from 20 to 19, then the first floating values ​​of other curves also move in the same way, also from 20 to 19, and are synchronous). The number of related flow data between the first floating value and the second floating value in each flow data fluctuation curve at this moment during the moving process is determined and calibrated as G i , and i = 1, 2, ..., n, where i represents different flow data fluctuation curves;

[0041] The several groups G generated at different times i Perform standard deviation processing to determine the standard value BZ. The standard deviation processing method is: i Perform mean processing and determine the mean Jz, using Determine the standard value BZ, complete a movement process of the first floating value and the second floating value (that is, the first floating value moves downward from the smallest highest value in several curves, and the second floating value moves upward from the largest lowest value in several curves, completing a movement process. When the movement is repeated, it means that a movement process is completed). Select the minimum value from several groups of standard values ​​BZ generated in this movement process, and determine the specific positions of the first floating value and the second floating value corresponding to the minimum value. Based on the relevant numerical values ​​F2 and F1 corresponding to the first floating value and the second floating value (F2 corresponds to the first floating value, and F1 corresponds to the second floating value), determine the medium flow change interval [F1, F2];

[0042] Select the minimum value Smin from the minimum flow data within the fluctuation curves of several flow data to determine the low flow variation interval [Smin, F1), and then select the minimum value Smax from the maximum flow data within the fluctuation curves of several flow data to determine the low flow variation interval (F2, Smax]);

[0043] The reason why two floating values ​​are used here is to confirm the traffic usage of different platforms or related web pages under different conditions, so that the determined traffic change interval is more accurate. The standard deviation can show the discreteness of the data. When the number of values ​​corresponding to the floating interval determined based on the floating process is the closest, then the data discreteness of this situation is the lowest, and the determined related interval is the most accurate. In order to facilitate the relevant confirmation of the subsequent push time period, it is necessary to determine the accuracy of this interval first.

[0044] Among them, the associated time period determination end determines the associated time periods included in the intervals from a number of traffic data fluctuation curves based on the determined high traffic change interval, medium traffic change interval and low traffic change interval, and determines the high time period, low time period and medium time period based on the intersection of the associated time periods in different change intervals (because the high traffic change interval is the situation where users normally use this platform, if information is pushed during this period, it will cause disgust to relevant users, so it is not considered. The reason why the high time period is determined is that it is necessary to identify the corresponding user's interest in this platform based on the proportion of the high time period in the entire 24 hours, so as to determine the relevant number of subsequent pushed information);

[0045] The specific method for determining the middle period is as follows:

[0046] Based on the determined medium flow variation interval, the associated time periods belonging to this variation interval are confirmed from several flow data fluctuation curves, and the associated time periods of different flow data fluctuation curves are cross-processed to determine the cross-over time period. Based on the number SG of associated time periods corresponding to the cross-over time period, a group of cross-over time periods with the largest number SG is determined as the medium time period. If there is a situation where the number SG is the largest and the same, then both groups of cross-over time periods are medium time periods. For example, the associated time periods are assumed to be A[2, 5], B[3, 6], and C[3, 6.5], then the cross-over time period generated by the intersection of the three time periods A, B, and C is [3, 5]. Then this time period is the determined medium time period, and the number of its associated time periods is 3. The cross-over time period [3, 6] generated by the intersection between C and B has only two corresponding associated time periods, and the number is 2, so it is not selected.

[0047] The specific method for determining the low period is:

[0048] Based on the determined low flow variation interval, the low time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period;

[0049] The specific method for determining the high period is:

[0050] Based on the determined high flow variation interval, the high time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period;

[0051] The confirmed high time period, medium time period and low time period are all transmitted to the push center.

[0052] Specifically, the reason why different relevant time periods are determined here is that the relevant time periods when the corresponding users use the corresponding platforms can further demonstrate the relevant usage habits of the corresponding users. In combination with the relevant time periods, different push logics are set in different time periods to push relevant information, thereby ensuring the accuracy of the information push and also ensuring that the effect of the information push can be better.

[0053] Among them, the push database has the relevant push information of the relevant platforms preset inside, which is set in advance by the relevant platform personnel;

[0054] The push center determines the number of push messages based on the relevant proportion of the high time period, and randomly selects from the push database, and then combines the current time. If the current time belongs to the low time period, the low time period push unit is used to push the push message. If the current time belongs to the medium time period, the medium time period push unit is used to push the push message. The specific execution processing method of the push center is as follows:

[0055] Determine the proportion of the high time period in 24 hours (based on the total duration of the high time period and the total duration of 24 hours, the relevant proportion can be confirmed), and compare the proportion with the preset interval, and there are multiple groups of preset intervals, and the endpoint values ​​of each group of preset intervals are preset in advance. Each group of preset intervals corresponds to a different number of pushes. Based on the preset interval to which the proportion belongs, determine the number of push information TG (TG is no more than 4), and select TG push information from the push database;

[0056] Determine the number of low periods between 0:00 and 24:00 (that is, the timeline of the day):

[0057] If the total number of low-frequency time periods is ≥TG, randomly select push information in the middle of different low-frequency time periods and push the information in sequence until the push is completed;

[0058] If the total number of low time periods is less than TG, a group of push information is selected in the middle time of each subsequent low time period, and when the corresponding low time period arrives, the push information is pushed directly, and the remaining push information is pushed through the subsequent middle time period, and the current time is confirmed, and the time to which the current time belongs is identified:

[0059] If the current time belongs to the low time period, the information will be pushed directly through the low time period push unit, and the time will be monitored. When the current time belongs to the medium time period, the remaining push information will be pushed through the medium time period push unit, and the push time interval will be set in advance, and the push time interval will start from the first information push;

[0060] If the current time belongs to the mid-time period, the remaining push information will be pushed through the mid-time push unit, and the push time interval is set in advance.

[0061] Specifically, the purpose of pushing information during low-time periods is to encourage relevant users to use the platform again and increase the usage rate of the platform. Because low-time periods are unused relevant periods, information push is required during each low-time period. This type of pushed information will not affect the user experience of relevant personnel, but will also enable relevant personnel to open the platform and achieve better information push effect.

[0062] Embodiment 2

[0063] Further implementation method for the current moment belonging to the middle period:

[0064] The time period push unit identifies whether the platform traffic data belongs to the medium traffic change interval based on the platform traffic data at the current moment. If it does not belong to the medium traffic change interval, no information push is performed. When the platform traffic data belongs to the medium traffic change interval and the current moment also belongs to the medium time period, information push is performed. If it belongs to the medium traffic change interval, information push is performed directly.

[0065] The reason for such limitation here is to ensure the accuracy of information pushed during the mid-time period. Because relevant personnel may watch this platform during the mid-time period, when information is pushed, relevant personnel may click on the pushed information to view it. This kind of click-through method has a better push effect. Therefore, when pushing during the mid-time period, it needs to be more accurate in order to ensure the relevant push effect.

[0066] Some of the data in the above formulas are numerically calculated by removing their dimensions. Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art.

[0067] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An artificial intelligence push system based on big data, characterized in that: include: The relevant data collection end collects the relevant flow data of different standard periods used by relevant personnel on the relevant platform, and transmits the collected relevant flow data of different standard periods to the relevant data processing end; The relevant data processing end processes the collected relevant flow data of each different standard period, generates the flow data fluctuation curve of the corresponding standard period according to the different relevant flow data corresponding to different moments in the same standard period, and then uniformly processes the generated groups of fluctuation curves to determine the high flow change interval, medium flow change interval and low flow change interval of the relevant platform. The specific method is as follows: Confirm several groups of related flow data in the same standard period, use the timeline as the horizontal coordinate axis and the flow value as the vertical coordinate axis, and generate the flow data fluctuation curve belonging to this standard period in the coordinate system based on the changes of several groups of related flow data in this standard period; Confirm the maximum flow data and the minimum flow data of the flow data fluctuation curve corresponding to the standard period, and randomly select the first floating value and the second floating value from the maximum flow data and the minimum flow data, wherein there is a limiting standard: the second floating value < the first floating value < the maximum flow data, the minimum flow data < the second floating value < the first floating value; The first floating value and the second floating value of each group of flow data fluctuation curve are confirmed one by one, so that the first floating value and the second floating value are moved up and down. The moving process cannot exceed the original limit standard, and the first floating value and the second floating value of each flow data fluctuation curve that move up and down are consistent. The number of related flow data between the first floating value and the second floating value in each flow data fluctuation curve at this moment during the moving process is determined and calibrated as G i , and i = 1, 2, ..., n, where i represents different flow data fluctuation curves; The several groups G generated at different times i Perform standard deviation processing to determine the standard value BZ, complete a movement process of the first floating value and the second floating value, select the minimum value from the several groups of standard values ​​BZ generated in this movement process, and determine the specific positions of the first floating value and the second floating value corresponding to the minimum value, and determine the medium flow change interval [F1, F2] based on the relevant values ​​F2 and F1 corresponding to the first floating value and the second floating value; Select the minimum value Smin from the minimum flow data within the fluctuation curves of several flow data to determine the low flow variation interval [Smin, F1), and then select the minimum value Smax from the maximum flow data within the fluctuation curves of several flow data to determine the low flow variation interval (F2, Smax]); The associated time period determination end determines the associated time periods included in the intervals from a number of flow data fluctuation curves based on the determined high flow variation interval, medium flow variation interval, and low flow variation interval, and determines the high time period, low time period, and medium time period based on the intersection of associated time periods in different variation intervals. The specific method is as follows: Based on the determined medium flow variation interval, the associated time periods belonging to the variation interval are confirmed from a number of flow data fluctuation curves, and the associated time periods of different flow data fluctuation curves are cross-processed to determine the cross-time periods. Based on the number SG of associated time periods corresponding to the cross-time periods, a group of cross-time periods with the largest number SG is determined as the medium time period. If there is a situation where the number SG is the largest and the same, then both groups of cross-time periods are medium time periods. The specific method for determining the low period is: Based on the determined low flow variation interval, the low time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period; The specific method for determining the high period is: Based on the determined high flow variation interval, the high time period is determined by adopting the same processing method of several flow data fluctuation curves when confirming the middle time period; The confirmed high time period, medium time period and low time period are transmitted to the push center; The push center determines the number of push messages based on the relevant proportion of the high-time period, and randomly selects from the push database. Combined with the current time, the low-time push unit or the medium-time push unit is used to push the push messages.

2. According to claim 1, an artificial intelligence push system based on big data is characterized in that: The standard period is 24 hours, and the relevant data collection terminal collects at least five sets of relevant flow data of the standard period.

3. According to claim 1, an artificial intelligence push system based on big data is characterized in that: The related data processing end performs unified processing on the generated groups of fluctuation curves.

4. According to claim 1, an artificial intelligence push system based on big data is characterized in that: The specific method of the push center to push the push information is as follows: Determine the proportion of high-frequency time periods within 24 hours, and compare the proportion with the preset intervals. There are multiple groups of preset intervals, and the endpoint values ​​of each group of preset intervals are preset in advance. Each group of preset intervals corresponds to a different number of pushes. Based on the preset interval to which the proportion belongs, determine the number of push information TG, and select TG push information from the push database; Determine the number of low periods between 0:00 and 24:00: If the total number of low-frequency time periods is ≥TG, randomly select push information in the middle of different low-frequency time periods and push the information in sequence until the push is completed; If the total number of low time periods is less than TG, a group of push information is selected in the middle time of each subsequent low time period, and when the corresponding low time period arrives, the push information is pushed directly, and the remaining push information is pushed through the subsequent middle time period, and the current time is confirmed, and the time to which the current time belongs is identified: If the current time belongs to the low time period, the information will be pushed directly through the low time period push unit, and the time will be monitored. When the current time belongs to the medium time period, the remaining push information will be pushed through the medium time period push unit, and the push time interval will be set in advance, and the push time interval will start from the first information push; If the current time belongs to the mid-time period, the remaining push information will be pushed through the mid-time push unit, and the push time interval is set in advance.

Images