A short video user click prediction method based on artificial intelligence big data

Abstract

The present invention discloses a short video user click prediction method based on artificial intelligence big data. Step 1: obtain the user's click behavior data based on different short video types, analyze the click behavior data, obtain different click behavior data attributes, and mark them as XW i ; Step 2: Determine the type of relevant short videos and mark them as SP q d ; Determine the user's click behavior on different videos of the same type and mark them, and calculate the user's liking value CD1 for the type of video through an algorithm; Step three, compare the calculated liking value CD1 with the preset value YS to complete the preliminary prediction of the user's click behavior. This method of predicting short video user clicks based on artificial intelligence big data obtains the attribute analysis of the user's different click behaviors, and assigns weight coefficient values ​​to the attributes of different click behaviors to calculate the user's liking value for the short video, thereby more accurately predicting the user's click behavior.

Description

Technical Field

[0001] The present invention belongs to the technical field of short video user click prediction, and in particular relates to a short video user click prediction method based on artificial intelligence big data. Background Art

[0002] Short videos are a form of Internet content dissemination. The quality varies, and the length of short videos is generally less than 5 minutes. With the popularization of mobile terminals and the acceleration of the Internet, they have gradually gained the favor of major platforms, fans, and capital. Predicting users' click behavior can accurately understand their preferences, so that relevant content can be recommended to users, improving their experience;

[0003] In the prior art, machine learning and data analysis are used to predict user click behaviors, which can bring better experience to users to a certain extent. However, according to big data statistics, there are still some users who often see short videos that they are not interested in. Only users can reduce the recommendation of related short videos by clicking on "not interested". Therefore, the prior art predicts user click behaviors, but there are still certain defects. In addition, the prior art does not specifically analyze the situation of inaccurate prediction. To this end, the present invention provides a short video user click prediction method based on artificial intelligence big data. Summary of the invention

[0004] The purpose of the present invention is to provide a short video user click prediction method based on artificial intelligence big data to solve the technical problems raised in the background technology.

[0005] To achieve the above purpose, the specific technical solution of the present invention is as follows: A short video user click prediction method based on artificial intelligence big data, specifically comprising the following steps:

[0006] Step 1: Obtain the user's click behavior data based on different short video types, analyze the click behavior data, obtain different click behavior data attributes, and mark them as XW i ;

[0007] Step 2: Determine the type of relevant short videos and mark them as SP q d ; Mark the user's click behavior on different videos of the same type, and calculate the user's preference value CD1 for the video of this type through an algorithm;

[0008] Step 3: Compare the calculated likeability value CD1 with the preset value YS to complete the preliminary prediction of the user's click behavior;

[0009] Step 4: Further judge the preliminary prediction results of click behavior according to the degree of preference for different types of short videos, and identify three types of users: normal users, users with higher click behavior than normal users, and users with lower click behavior than normal users;

[0010] For normal users, the preliminary prediction result of click behavior in step 3 is directly used as the prediction result for the user;

[0011] For users whose click behavior is higher than normal, the likeability values ​​of different short videos are arranged in a ranking manner to complete the prediction of such users;

[0012] For users whose click behavior is lower than normal, the user's preference value SCD1 for short videos based on time is calculated to predict their click behavior.

[0013] Preferably, the specific method of preliminary prediction of click behavior is as follows:

[0014] S1. Get different click behavior data attributes and mark them as XW i , i represents different behavior data attributes, and i=1, 2...n, n represents the total number of different behavior data attributes;

[0015] S2. Determine the type of relevant short videos and mark them as SP q d , q represents different types of short videos, and q=1, 2, ...m, m represents the total number of short videos of different types; d represents different short videos of the same type, and d=1, 2, ...g, g represents the total number of different short videos of the same type;

[0016] S3. When q is 1, the click behavior mark of the user on different videos of the same type is {(XW1, SP 1 d ), (XW2, SP 1 d ), (XW3, SP 1 d )......(XW n , S.P. 1 d )}, and count the number of different user behavior attributes of g videos of the same type, and determine XW1, XW1, ... XW respectively. i The number of are X1, X2, ...Xi, and according to the number of different behavior attributes, the user's preference value CD1 for this type of video is calculated through an algorithm. The specific algorithm is as follows:

[0017]

[0018] Among them, P1, P2, ...Pn are weight coefficient values ​​of different behavior data attributes. Different click behaviors have different likes for short videos. The higher the like, the greater the weight coefficient value assigned to the corresponding click behavior data attribute;

[0019] The calculated liking value CD1 is compared with the preset value YS. YS is a standard value set artificially. When CD1>YS, it means that the user likes this type of video highly. When CD1<YS, it means that the user likes this type of video generally or does not like it. This completes the preliminary prediction of the user's click behavior.

[0020] Preferably, the specific determination methods of the three user types are as follows:

[0021] T1. Find the likeness value of different types of short videos, that is, when q is 1, 2, 3...m, find CD respectively. 1、 CD 2、 CD3.....CD m ;

[0022] T2. Use the likeability values ​​of different types of videos as the X-axis and the likeability values ​​as the Y-axis to establish a two-dimensional coordinate axis and determine CD 1、 CD 2、 CD3.....CD m Find the point on the two-dimensional coordinate axis and find the YS value. Draw the first reference line L1 and the second reference line L2 in the X-axis direction with YS+1 and YS-1 respectively.

[0023] The area between T3, L1 and L2 is the standard threshold, and CD is selected 1、 CD 2、 CD3.....CD m The highest point and the lowest point in the image are connected to determine the middle value and the reference line L3 is determined along the X-axis;

[0024] If L3 is between L1 and L2, it means that the user has normal click behavior, which means that the initial prediction of click behavior is highly accurate;

[0025] If L3 is above L1, it means that the user's click behavior on all types of short videos is higher than that of normal users, which means that the initial prediction of click behavior is inaccurate and insufficient to express the user's liking for the video.

[0026] If L3 is below L1: it means that the user's click behavior on all types of short videos is lower than that of normal users, or even no click behavior, so the initial prediction of click behavior is inaccurate, and further analysis is needed on the user's non-click behavior.

[0027] Preferably, the click behavior prediction method for users with higher click behavior than normal users is as follows: the likeness values ​​of different short videos are arranged in descending order, and the short video corresponding to the type of the first likeness value is the user's favorite, and {CD 1、 CD 2、 CD3.....CD m}Just arrange them in order from largest to smallest.

[0028] Preferably, the click behavior prediction method for users whose click behavior is lower than normal is as follows:

[0029] P1. Determine the short video SP of the same type d , and determine the playing time t for the short video d , {t1, t2, ... t g} and determine the viewing time T for short videos d {T1, T2, ... T g}, d represents different short videos of the same type, and d=1, 2, ... g, g represents the total number of different short videos of the same type,

[0030] P2, and calculate the time difference △t of each video watched by the user g , {△t1,△t2,......△t g},△t g =T g -t g ; △t g May be negative, when △t g <0, then delete it;

[0031] Based on the calculated time difference, determine the user's preference value SCD1 for the short video based on time, and select △t greater than 0 g The following calculation is performed, and the specific algorithm is as follows:

[0032]

[0033] Where k is the time coefficient, which is a manually set value;

[0034] P3. Compare the calculated SCD1 with the preset time value SYS. SYS is a standard value set artificially. When SCD1>SYS, it means that the user likes this type of video very much. When SCD1<SYS, it means that the user likes this type of video in general or does not like it, thus completing the prediction of the user's click behavior.

[0035] The short video user click prediction method based on artificial intelligence big data of the present invention has the following advantages: the short video user click prediction method based on artificial intelligence big data obtains the user's attribute analysis of different click behaviors and assigns weight coefficient values ​​to different click behavior attributes to calculate the user's liking degree for short videos, thereby more accurately predicting the user's click behavior; on this basis, different user types are judged separately to distinguish three user types: normal users, users with higher click behaviors than normal users, and users with lower click behaviors than normal users; and the user types with higher click behaviors than normal users and users with lower click behaviors are judged separately by the liking degree value sorting method and the short video viewing time judgment method, thereby completing a more accurate prediction of the user's click behavior, thereby comprehensively promoting different user types to enable users to have a better experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 It is the overall flow chart of the present invention;

[0038] Figure 2 It is a two-dimensional coordinate diagram for determining the user type in the present invention. DETAILED DESCRIPTION

[0039] In the following, only some exemplary embodiments are briefly described. As those skilled in the art will appreciate, the described embodiments may be modified in various ways without departing from the spirit or scope of the embodiments of the present invention. Therefore, the drawings and descriptions are considered to be exemplary and non-restrictive in nature.

[0040] In the description of the embodiments of the present invention, it should be understood that the terms "length", "vertical", "horizontal", "top", "bottom", etc. indicating the orientation or positional relationship are based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present invention.

[0041] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the embodiments of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0042] In the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection, an electrical connection, or a communication; it can be a direct connection, or an indirect connection through an intermediate medium, and it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in the embodiments of the present invention can be understood according to specific circumstances.

[0043] The disclosure below provides many different embodiments or examples to implement different structures of the embodiments of the present invention. In order to simplify the disclosure of the embodiments of the present invention, the parts and settings of specific examples are described below. Of course, they are only examples, and the purpose is not to limit the embodiments of the present invention. In addition, the embodiments of the present invention can repeat reference numbers and / or reference letters in different examples, and this repetition is for the purpose of simplification and clarity, which itself does not indicate the relationship between the various embodiments and / or settings discussed.

[0044] In order to better understand the purpose, structure and function of the present invention, the following is a further detailed description of a short video user click prediction method based on artificial intelligence big data in conjunction with the accompanying drawings.

[0045] like Figure 1-2 As shown, a short video user click prediction method based on artificial intelligence big data of the present invention includes the following embodiments:

[0046] Example 1

[0047] S1. Obtain the user's click behavior data based on different short video types, analyze the click behavior data, obtain different click behavior data attributes, and mark them as XW i , i represents different behavior data attributes, and i=1, 2...n, n represents the total number of different behavior data attributes;

[0048] S2. Determine the type of relevant short videos and mark them as SP q d, q represents different types of short videos, and q=1, 2, ...m, m represents the total number of short videos of different types; d represents different short videos of the same type, and d=1, 2, ...g, g represents the total number of different short videos of the same type;

[0049] S3. When q is 1, the click behavior mark of the user on different videos of the same type is {(XW1, SP 1 d ), (XW2, SP 1 d ), (XW3, SP 1 d )......(XW n , S.P. 1 d )}, and count the number of different user behavior attributes of g videos of the same type, and determine XW1, XW1, ... XW respectively. i The number of are X1, X2, ...Xi, and according to the number of different behavior attributes, the user's preference value CD1 for this type of video is calculated through an algorithm. The specific algorithm is as follows:

[0050]

[0051] Among them, P1, P2, ...Pn are weight coefficient values ​​of different behavior data attributes respectively. Different click behaviors have different likes for short videos. The higher the like, the greater the weight coefficient value assigned to the corresponding click behavior data attribute; (For example, click behaviors can be: like, comment, forward, favorite, follow. Obviously, the higher the like of the short video for the favorite and follow behaviors, the greater the weight coefficient value of the corresponding behavior data attribute);

[0052] S4. Compare the calculated liking value CD1 with the preset value YS. YS is a standard value set artificially. When CD1>YS, it means that the user likes this type of video highly. When CD1<YS, it means that the user likes this type of video generally or does not like it. This completes the preliminary prediction of the user's click behavior.

[0053] Example 2

[0054] T1. Find the likeness value of different types of short videos, that is, when q is 1, 2, 3...m, find CD respectively. 1、 CD 2、 CD3.....CD m ;

[0055] T2. Use the likeability values ​​of different types of videos as the X-axis and the likeability values ​​as the Y-axis to establish a two-dimensional coordinate axis and determine CD 1、 CD 2、 CD3.....CD m Find the point on the two-dimensional coordinate axis and find the YS value. Draw the first reference line L1 and the second reference line L2 in the X-axis direction with YS+1 and YS-1 respectively.

[0056] The area between T3, L1 and L2 is the standard threshold, and CD is selected 1、 CD 2、 CD3.....CD m The highest point and the lowest point in the image are connected to determine the middle value and the reference line L3 is determined along the X-axis;

[0057] If L3 is between L1 and L2, it means that the user has normal click behavior, which means that the initial prediction of click behavior is highly accurate;

[0058] If L3 is above L1, it means that the user's click behavior on all types of short videos is higher than that of normal users, which means that the initial prediction of click behavior is inaccurate and insufficient to express the user's liking for the video. At this time, the liking values ​​of different short videos can be arranged in descending order. 1、 CD 2、 CD3.....CD m}Just arrange them in descending order, and the short video corresponding to the type that ranks first in popularity value is the one that users like to watch the most.

[0059] If L3 is below L1: it means that the user's click behavior on all types of short videos is lower than that of normal users, or even no click behavior, so the initial prediction of click behavior is inaccurate, and further analysis is needed on the user's non-click behavior.

[0060] Example 3

[0061] P1. Determine the short video SP of the same type d , and determine the playing time t for the short video d , {t1, t2, ... t g} and determine the viewing time T for short videos d {T1, T2, ... T g}, d represents different short videos of the same type, and d=1, 2, ... g, g represents the total number of different short videos of the same type,

[0062] P2, and calculate the time difference △t of each video watched by the user g , {△t1,△t2,......△tg},△t g =T g -t g ; △t g May be negative, when △t g <0, delete it;

[0063] Based on the calculated time difference, determine the user's preference value SCD1 for the short video based on time, and select △t greater than 0 g The following calculation is performed, and the specific algorithm is as follows:

[0064]

[0065] Where k is the time coefficient, which is a manually set value.

[0066] P3. Compare the calculated SCD1 with the preset time value SYS. SYS is a standard value set artificially. When SCD1>SYS, it means that the user likes this type of video very much. When SCD1<SYS, it means that the user likes this type of video in general or does not like it, thus completing the prediction of the user's click behavior.

[0067] The working principle of this short video user click prediction method based on artificial intelligence big data:

[0068] It is to be understood that the present invention is described by some embodiments, and it is known to those skilled in the art that various changes or equivalent substitutions may be made to these features and embodiments without departing from the spirit and scope of the present invention. In addition, under the teachings of the present invention, these features and embodiments may be modified to adapt to specific circumstances and materials without departing from the spirit and scope of the present invention. Therefore, the present invention is not limited by the specific embodiments disclosed herein, and all embodiments falling within the scope of the claims of this application are within the scope of protection of the present invention.

Claims

1. A short video user click prediction method based on artificial intelligence big data, characterized by: The specific steps include: Step 1: Obtain the user's click behavior data based on different short video types, analyze the click behavior data, obtain different click behavior data attributes, and mark them as XW i ; Step 2: Determine the type of relevant short videos and mark them as SP q d ; Determine the user's click behavior on different videos of the same type and mark them, and calculate the user's preference value CD1 for videos of this type through an algorithm; Step 3: Compare the calculated likeability value CD1 with the preset value YS to complete the preliminary prediction of the user's click behavior; Step 4: Further judge the preliminary prediction results of click behavior according to the degree of preference for different types of short videos, and identify three types of users: normal users, users with higher click behavior than normal users, and users with lower click behavior than normal users; For normal users, the preliminary prediction result of click behavior in step 3 is directly used as the prediction result for the user; For users whose click behavior is higher than normal, the likeability values ​​of different short videos are arranged in a ranking manner to complete the prediction of such users; For users whose click behavior is lower than normal, find the user's preference value SCD1 based on time for the short video to predict the click behavior; The specific method of preliminary prediction of click behavior is as follows: S1. Get different click behavior data attributes and mark them as XW i , i represents different behavior data attributes, and i=1, 2...n, n represents the total number of different behavior data attributes; S2. Determine the type of relevant short videos and mark them as SP q d , q represents different types of short videos, and q=1, 2, ...m, m represents the total number of short videos of different types; d represents different short videos of the same type, and d=1, 2, ...g, g represents the total number of different short videos of the same type; S3. When q is 1, the click behavior mark of the user on different videos of the same type is {(XW1, SP 1 d ), (XW2, SP 1 d ), (XW3, SP 1 d )......(XW n , S.P. 1 d )}, and count the number of different user behavior attributes of g videos of the same type, and determine XW1, XW1, ... XW respectively. i The number of are X1, X2, ...Xi, and according to the number of different behavior attributes, the user's preference value CD1 for this type of video is calculated through an algorithm. The specific algorithm is as follows: Among them, P1, P2, ...Pn are weight coefficient values ​​of different behavior data attributes. Different click behaviors have different likes for short videos. The higher the like, the greater the weight coefficient value assigned to the corresponding click behavior data attribute; Compare the calculated likeness value CD1 with the preset value YS. YS is a standard value set manually. When CD1>YS, it means that the user likes this type of video highly. When CD1<YS, it means that the user likes this type of video generally or does not like it. This completes the preliminary prediction of the user's click behavior. The specific judgment methods of the three user types are as follows: T1. Find the likeness value of different types of short videos, that is, when q is 1, 2, 3...m, find CD respectively. 1、 CD 2、 CD3......CD m ; T2. Use the likeability values ​​of different types of videos as the X-axis and the likeability values ​​as the Y-axis to establish a two-dimensional coordinate axis and determine CD 1、 CD 2、 CD3......CD m Find the point on the two-dimensional coordinate axis and find the YS value. Draw the first reference line L1 and the second reference line L2 in the X-axis direction with YS+1 and YS-1 respectively. The area between T3, L1 and L2 is the standard threshold, and CD is selected 1、 CD 2、 CD3......CD m The highest point and the lowest point in the image are connected to determine the middle value and the reference line L3 is determined along the X-axis; If L3 is between L1 and L2, it means that the user has normal click behavior, which means that the initial prediction of click behavior is highly accurate; If L3 is above L1, it means that the user's click behavior on all types of short videos is higher than that of normal users, which means that the initial prediction of click behavior is inaccurate and insufficient to express the user's liking for the video; If L3 is below L1: it means that the user's click behavior on all types of short videos is lower than that of normal users, or even no click behavior, so the initial prediction of click behavior is inaccurate, and further analysis is needed on the user's non-click behavior.

2. The method for predicting short video user clicks based on artificial intelligence big data according to claim 1, characterized in that: The prediction method for the click behavior of users with higher click behavior than normal users is as follows: the likeability values ​​of different short videos are arranged in descending order, and the short video corresponding to the type of the first likeability value is the user's favorite. 1、 CD 2、 CD3......CD m }Just arrange them in order from largest to smallest.

3. The method for predicting short video user clicks based on artificial intelligence big data according to claim 1, characterized in that: The click behavior prediction method for users whose click behavior is lower than normal is as follows: P1. Determine the short video SP of the same type d , and determine the playing time t for the short video d , {t1, t2, ... t g } and determine the viewing time T for short videos d {T1, T2, ... T g }, d represents different short videos of the same type, and d=1, 2, ... g, g represents the total number of different short videos of the same type, P2, and calculate the time difference △t of each video watched by the user g , {△t1,△t2,......△t g },△t g =T g -t g ; △t g May be negative, when △t g <0, delete it; Based on the calculated time difference, determine the user's preference value SCD1 for the short video based on time, and select △t greater than 0 g The following calculation is performed, and the specific algorithm is as follows: Where k is the time coefficient, which is a manually set value; P3. Compare the calculated SCD1 with the preset time value SYS. SYS is a standard value set artificially. When SCD1>SYS, it means that the user likes this type of video very much. When SCD1<SYS, it means that the user likes this type of video in general or does not like it, thus completing the prediction of the user's click behavior.

Images