Operation strategy generation method based on account data analysis

Abstract

The application relates to the technical field of short video operation, and discloses an operation strategy generation method based on account data analysis, which comprises the following steps: when a preset time period is reached, according to the account information of a user on each short video platform, short video browsing record information corresponding to the account information is acquired; short video type identification is performed on the short video browsing record information, and a first short video type set is obtained; the actual proportion values of each short video type are counted; according to the actual proportion values, a preset proportion threshold value stored in advance is called, short video types with actual proportion values less than the preset proportion threshold value are removed, a second short video type set is generated; the total proportion value of the second short video type set is calculated, an adjustment strategy stored in a database in advance is called, and a short video recommendation strategy of the user is determined; according to the short video recommendation strategy of the user and short videos corresponding to each short video type stored in the database, an optimal short video recommendation scheme of the user at the current moment is generated based on an improved genetic algorithm.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of short video operation, in particular to an operation strategy generation method based on account data analysis. BACKGROUND

[0002] Mobile short video applications represented by Kuaishou and Tiktok are a new content dissemination media after text, pictures and traditional videos. In recent years, mobile short video applications have developed rapidly and are becoming a new social platform and super portal in the mobile internet era. Since the user's sensory experience of short videos mainly reflects in the process of short video browsing, the recommended short videos on the platform, therefore, it is crucial to make reasonable short video recommendations to users.

[0003] Existing short video recommendation methods only consider a single platform, with little data connection between platforms. When recommending short videos, they are recommended based on the browsing data of their own short videos. When it is identified that a user is interested in a certain type of short video at a certain time period, the same type of short video will be recommended for a long time, which will lead to poor overall user sensory experience and greatly reduce user experience. SUMMARY

[0004] One of the purposes of the present application is to provide an operation strategy generation method based on account data analysis, which can produce the optimal solution for short video recommendation, thereby providing the best short video browsing experience for users and greatly improving user experience.

[0005] In order to achieve the above purpose, an operation strategy generation method based on account data analysis is provided, comprising the following steps:

[0006] S1. When a preset time period is reached, the short video browsing record information corresponding to each account information is obtained according to the account information of the user on each short video platform;

[0007] S2. Based on the short video browsing record information of all account information of the user, the short video type identification is performed on the short video browsing record information to obtain the corresponding first short video type set, and the actual proportion value of each identified short video type is counted;

[0008] S3. According to the actual proportion value of the user's short video type after statistical processing, the pre-stored preset proportion threshold value in the database is retrieved, the short video type with actual proportion value less than the preset proportion threshold value is eliminated, and the corresponding second short video type set is generated;

[0009] S4, according to the actual proportion value corresponding to the second short video type set, the total proportion value corresponding to the second short video type set is calculated, the adjustment strategy stored in the database is called to determine the short video recommendation strategy of the user; the short video recommendation strategy is the proportion between the short video type in the second short video type set and other short video types in a preset time period;

[0010] S5, according to the determined short video recommendation strategy of the user and the short video corresponding to each short video type stored in the database, the improved genetic algorithm is used to generate the optimal short video recommendation scheme corresponding to the current time of the user.

[0011] The principle and effect of the scheme: first, after reaching the preset time period, the account information of the user on each short video platform is obtained, and the corresponding short video browsing record information is obtained through the account information, so that the short video browsing record of the user can be viewed and counted. The account information on each short video platform is used to count and identify the user's preferences, which greatly improves the accuracy and comprehensiveness of the user statistics.

[0012] Then, based on the short video browsing record information, the browsing proportion value of each short video type in the browsing process of the user is calculated, so as to obtain the first short video type set. The actual proportion value identified and the preset proportion threshold are used to screen the first short video type set to realize a more accurate second short video type set, which makes the data more accurate and the corresponding data validity is improved. Then, according to the total proportion value of the second short video type set, the adjustment strategy is selected, and then according to the short video recommendation strategy and each short video, the improved genetic algorithm is used to make the optimal short video recommendation scheme, so as to give the user the best short video browsing experience and greatly improve the user experience.

[0013] Further, the S5 includes the following steps:

[0014] S50, according to the determined short video recommendation strategy of the user, the corresponding other short video type and the short video type in the second short video type set are selected to determine the business set of the optimal short video type arrangement plan of the user, and the short video corresponding to each short video type is called from the database to form a new short video arrangement plan business set;

[0015] S51, determine the constraint condition corresponding to each business in the business set;

[0016] S52, establish the target function corresponding to each business in the business set;

[0017] S53, based on the improved genetic algorithm, and the determined objective function and constraint condition, multi-objective optimization of the short video type arrangement planning in the service set is realized;

[0018] S54, according to the result of multi-objective optimization, the short video type arrangement planning scheme of the service set is optimized, and the corresponding optimal short video recommendation scheme is generated.

[0019] Beneficial effect: in the scheme, first, the arrangement of the short video type is carried out according to the short video recommendation strategy, then the arrangement of the corresponding short video is carried out according to the type arrangement, so as to realize the arrangement of the short video corresponding to each short video type, so as to obtain the short video arrangement scheme based on the short video type, and then the constraint condition and the objective function are determined according to the service set of the short video arrangement, so as to realize the multi-objective optimization of the short video arrangement planning in the service set based on the improved genetic algorithm.

[0020] Further, the S53 comprises the following steps:

[0021] S530, an initial population with a scale of N is randomly generated, the individual of the initial population is the arrangement planning path of the short video corresponding to each short video type in the current time period, and the individual of the initial population is screened through the constraint condition, the constraint condition includes the total time length constraint condition of the short video, the number of short video type continuous repetition constraint condition and the repeated short video constraint condition, if the constraint condition is met, the corresponding arrangement planning path becomes a feasible solution, otherwise it is an infeasible solution;

[0022] S531, the first fitness and the second fitness of the screened population are calculated respectively; the first fitness is calculated as follows:

[0023]

[0024] f1=D1

[0025] The D1 is the total value of the initial population, the X i is the value of the short video i itself, and f1 is the first fitness;

[0026] The second fitness is calculated as follows:

[0027]

[0028] f2=D2

[0029] The D2 is the total value of the initial population, the Y j is the short video viewing effective degree prediction value of the short video j when browsing, and f2 is the second fitness;

[0030] S532, within the first preset number of iterations, according to the first fitness corresponding to the population, select the population with the first fitness greater than or equal to the first fitness threshold value, and select the first five populations with the second fitness maximum in the population eliminated at this time, and save them to the backup library;

[0031] When the first preset number of iterations is exceeded, the population at this time is centralized with the population in the backup library to form a new population, and the population with the second fitness greater than or equal to the preset second fitness threshold value corresponding to the population at this time is selected;

[0032] S533, the selected population is obtained by genetic algorithm crossing and mutation to obtain offspring population;

[0033] S534, after obtaining the offspring population, continue to execute S531 until the preset iteration amount is met;

[0034] S535, output the offspring population as the optimal solution set of multi-objective optimization.

[0035] Beneficial effect: in the scheme, first, through the constraint condition, that is, the total length constraint condition of the short video, the short video type continuous repetition number constraint condition and the repeated short video constraint condition, the individuals in the initial population are screened, which realizes the screening of the huge initial population and reduces the data processing difficulty of the subsequent optimal scheme selection.

[0036] The calculation of the fitness of the population when selecting the population includes the total value of the short video value and the effective degree prediction, compared with the existing genetic algorithm using existing data to generate the optimal scheme, the scheme innovatively introduces the prediction mechanism on the basis of the genetic algorithm, that is, after arranging the corresponding short video, it is predicted whether the user will effectively watch the short video when browsing to the short video, so that the short video can be browsed by the user as much as possible, greatly improving the effectiveness and reliability of the short video recommendation, and giving the customer a better short video watching experience.

[0037] Further, the following steps are included:

[0038] S6, according to the account information corresponding to each user and the corresponding second short video type set, the fan portrait corresponding to each short video type is counted, and the fan portrait includes the audience group corresponding to the short video type, the viewing time distribution information, and the distribution of the audience group on each short video platform;

[0039] S7, according to the fan portrait counted, generate a short video operation strategy corresponding to each short video type, and push it to the short video operator corresponding to each short video type, and the short video operator makes the short video of the short video type according to the short video operation strategy.

[0040] Beneficial effects: in this scheme, the generation of short video operation strategy is carried out through the analysis of the fan portrait of each customer, which greatly improves the effectiveness and convenience of short video production, and the generated short video has higher acceptance, which is more conducive to the users who like to watch.

[0041] Further, the S7 comprises the following steps:

[0042] S70, according to the audience group corresponding to the short video type and the distribution of the audience group in each short video platform, determining the video operator corresponding to the short video type in the short video platform with the most audience group, and obtaining the first comment information of the audience group in all short videos corresponding to the video operator, and the second comment information of the non-audience group in all short videos corresponding to the video operator in the short video platform with the least audience group;

[0043] S71, according to the first comment information and the second comment information, identifying and screening the first comment information and the second comment information, judging whether the first comment information and the second comment information are suggestion information, if yes, retaining, otherwise rejecting;

[0044] S72, according to the current retained first comment information and second comment information, and the historical comment information retrieved from the database, determining the modification report of the short video type corresponding to the video operator, and forming the corresponding short video operation strategy;

[0045] S73, after forming the short video operation strategy, the short video operation strategy is pushed to the first time, according to the short video operation strategy, the short video operator makes the short video of the short video type. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 The logic block diagram of the operation strategy generation method based on account data analysis in embodiment one of the application. DETAILED DESCRIPTION

[0047] The following will be further described in detail through specific embodiments:

[0048] Embodiment one

[0049] A kind of operation strategy generation method based on account data analysis, as shown in Figure 1 , including the following steps:

[0050] S1, when reaching the preset time period, according to the account information of the user on each short video platform, the short video browsing record information corresponding to each account information is obtained; in the embodiment, the account information includes account nickname, associated mobile phone number, user name corresponding to the account, age, gender. The short video browsing record information includes the short video browsed and the browsing time. In the embodiment, the account information and the short video browsing record information of the user on each video platform are adopted, so as to realize comprehensive collection of the record of the user browsing the short video, fully considering the case that the user browses the short video on multiple short video platforms.

[0051] S2, based on the short video browsing record information of all account information of the user, the short video type identification is performed on the short video browsing record information, and the corresponding first short video type set is obtained; and the actual proportion value of each identified short video type is counted; in the embodiment, the actual proportion value of each short video type browsed by the user, such as the actual proportion value of the funny type of user A is 0.42, the actual proportion value of the game type is 0.23, the actual proportion value of the dance type is 0.13, and the actual proportion value of the song type is 0.22, is obtained by identifying and counting the short video browsing record information of the user.

[0052] S3, according to the actual proportion value of the short video type of the user counted, the preset proportion threshold value stored in the database in advance is called, the short video type with the actual proportion value less than the preset proportion threshold value is removed, and the corresponding second short video type set is generated; in the embodiment, it is considered that the user may browse a certain type of short video in the process of browsing the short video type, but only for a certain short video, and is not interested in other short videos of the short video type, so further screening is needed to remove some invalid short video types not interested by the user.

[0053] S4, according to the actual proportion value corresponding to the second short video type set, the total proportion value corresponding to the second short video type set is calculated, the adjustment strategy stored in the database in advance is called, and the short video recommendation strategy of the user is determined; the short video recommendation strategy is the ratio between the short video type in the second short video type set and other short video types in the preset time period;

[0054] S5, according to the determined short video recommendation strategy of the user and the short video corresponding to each short video type stored in the database, the improved genetic algorithm is used to generate the optimal short video recommendation scheme corresponding to the current time of the user.

[0055] The S5 includes the following steps:

[0056] S50, according to the determined short video recommendation strategy of the user, selecting corresponding other short video types and short video types in the second short video type set, determining a business set of the optimal short video type arrangement plan of the user, and calling short videos corresponding to each short video type from the database to form a new business set of the short video arrangement plan;

[0057] S51, determining the constraint conditions corresponding to each business in the business set;

[0058] S52, establishing the objective function corresponding to each business in the business set;

[0059] S53, based on the improved genetic algorithm and the determined objective function and constraint conditions, realizing multi-objective optimization of the short video type arrangement plan in the business set;

[0060] The S53 includes the following steps:

[0061] S530, randomly generating an initial population with a scale of N, the individual of the initial population being an arrangement planning path of the short video corresponding to each short video type in the current time period, and screening the individual of the initial population through the constraint conditions, the constraint conditions including a total short video time constraint condition, a short video type continuous repetition number constraint condition and a repeated short video constraint condition, if the constraint conditions are met, the corresponding arrangement planning path becomes a feasible solution, otherwise it is an infeasible solution; in this embodiment, each constraint condition is specifically: the total short video time is less than or equal to the preset time period, the short video type continuous repetition number is less than 4, and the repeated short video does not exist.

[0062] S531, calculating the first fitness and the second fitness of the screened population respectively; the first fitness is calculated as follows:

[0063]

[0064] f1=D1

[0065] The D1 is the total value of the initial population, the X i is the value of the short video itself, and f1 is the first fitness; in this embodiment, the value of the short video itself is mainly embodied by the short video type corresponding to the short video and the indicators such as the like number, the forwarding number and the collection number of the short video, and the value of the single short video is calculated through the weight value corresponding to each indicator.

[0066] The second fitness is calculated as follows:

[0067]

[0068] f2=D2

[0069] The D2 is the total value of the short video viewing effectiveness prediction in the initial population, Y n is the short video viewing effectiveness prediction value of the short video n when the user browses, and f2 is the second fitness; in this embodiment, the calculation of the short video viewing effectiveness prediction value of the short video when the user browses is mainly performed by the BP neural network model, specifically, the browsing habit information of the user, the short video like number, the forwarding number, and the collection number of the short video are obtained as inputs, so as to output the corresponding prediction value of the user who may watch, in this way, the prediction value is more accurate and accurate, and the accuracy and reliability of the prediction value are greatly improved.

[0070] S532, within the first preset iteration number, according to the first fitness corresponding to the population, select the population whose first fitness is greater than or equal to the first fitness threshold, and select the first five populations with the largest second fitness from the population removed at this time, and save them to the standby library;

[0071] When the first preset iteration number is exceeded, the population at this time is centralized with the population in the standby library to form a new population, and the population whose second fitness is greater than or equal to the preset second fitness threshold is selected;

[0072] S533, the selected population is obtained by the cross and mutation of the genetic algorithm to obtain the offspring population;

[0073] S534, after obtaining the offspring population, continue to execute S531 until the preset iteration amount is met;

[0074] S535, output the offspring population as the optimal solution set of multi-objective optimization.

[0075] S54, according to the result of multi-objective optimization, optimizing the short video type arrangement planning scheme of the business set to generate the corresponding optimal short video recommendation scheme.

[0076] Further comprising the following steps:

[0077] S6, according to the account information corresponding to each user and the corresponding second short video type set, the fan portrait corresponding to each short video type is counted, and the fan portrait includes the audience group corresponding to the short video type, the viewing time distribution information, and the distribution of the audience group on each short video platform;

[0078] S7, according to the counted fan portrait, generating a short video operation strategy corresponding to each short video type, and pushing it to the short video operator corresponding to each short video type, and the short video operator makes the short video of the short video type according to the short video operation strategy.

[0079] S7 includes the following steps:

[0080] S70. Based on the audience corresponding to the short video type and the distribution of the audience across various short video platforms, determine the video operator corresponding to the short video type on the short video platform with the largest audience, and obtain the first comment information of the audience in all short videos corresponding to the video operator, as well as the second comment information of non-audience members in all short videos corresponding to the video operator on the short video platform with the fewest audience members. In this embodiment, the first comment information is the comments of users who like the short video, while the second comment information is the comments of users who dislike the short video. By obtaining these two data and processing them, the short videos updated later can be more liked by users who like them, and at the same time, they can further attract users who dislike them to like them.

[0081] S71. Based on the first comment information and the second comment information, identify and filter the first comment information and the second comment information, and determine whether the first comment information and the second comment information are suggestions. If so, retain them; otherwise, remove them.

[0082] S72. Based on the currently retained first and second comment information, and by retrieving historical comment information from the database, determine the modification report for the short video type corresponding to the video operator, and formulate the corresponding short video operation strategy;

[0083] S73. After formulating a short video operation strategy, it is immediately pushed to the short video operation strategy team. Based on the strategy, the short video operator produces short videos of that type. In this embodiment, the fan profile includes the target audience for that short video type, viewing time distribution information, and the distribution of the target audience across various short video platforms.

[0084] The above-mentioned are only embodiments of the present application, and the common knowledge of the specific structure and characteristics in the scheme is described too much, the ordinary skilled in the art knows all the ordinary technical knowledge in the field of the present application before the application date or the priority date, can know all the prior art in the field, and has the ability to apply the conventional experimental means before the date, the ordinary skilled in the art can perfect and implement the scheme under the enlightenment given by the present application combined with their own ability, some typical known structure or known method should not become the obstacle for the ordinary skilled in the art to implement the present application. It should be pointed out that for those skilled in the art, without departing from the structure of the present application, a number of modifications and improvements can also be made, which should also be considered as the protection scope of the present application, which will not affect the effect and practicality of the patent. The protection scope of the present application should be subject to the content of its claims, and the specific implementation mode and the like in the specification can be used to explain the content of the claims.

Claims

1.A method for generating an operation strategy based on account data analysis, characterized in that: The method comprises the following steps: S1, when reaching a preset time period, obtaining short video browsing record information corresponding to each account information according to the account information of the user on each short video platform; S2, performing short video type identification on the short video browsing record information of all account information of the user, obtaining a corresponding first short video type set, and counting the actual proportion value of each identified short video type; S3, according to the actual proportion value of the short video type of the user counted, calling a preset proportion threshold value stored in the database, eliminating short video types with an actual proportion value less than the preset proportion threshold value, and generating a corresponding second short video type set; S4, according to the actual proportion value corresponding to the second short video type set, calculating the total proportion value corresponding to the second short video type set, calling the adjustment strategy stored in the database, and determining the short video recommendation strategy of the user; the short video recommendation strategy is the ratio between the short video types in the second short video type set and other short video types within the preset time period; S5, according to the determined short video recommendation strategy of the user and the short video corresponding to each short video type stored in the database, generating an optimal short video recommendation scheme corresponding to the current time of the user based on the improved genetic algorithm. 2.The method of claim 1, wherein: The S5 comprises the following steps: S50, according to the determined short video recommendation strategy of the user, selecting the corresponding other short video types and the short video types in the second short video type set, determining the business set of the optimal short video type arrangement plan of the user, and calling the short video corresponding to each short video type from the database to form a new short video arrangement plan business set; S51, determining the constraint condition corresponding to each business in the business set; S52, establishing the objective function corresponding to each business in the business set; S53, based on the improved genetic algorithm and the determined objective function and constraint condition, realizing multi-objective optimization of the short video type arrangement plan in the business set; S54, according to the result of multi-objective optimization, optimizing the short video type arrangement plan scheme of the business set to generate a corresponding optimal short video recommendation scheme. 3.The method of claim 2, wherein: The S53 comprises the following steps: S530, randomly generating an initial population with a scale of N, the individual of the initial population being the arrangement planning path of the short video corresponding to each short video type in the current time period, and screening the individual of the initial population through the constraint condition, the constraint condition including the short video total time constraint condition, the short video type continuous repetition number constraint condition and the repeated short video constraint condition, if the constraint condition is met, the corresponding arrangement planning path becomes a feasible solution, otherwise it is an infeasible solution; S531, calculating the first fitness and the second fitness of the screened population respectively; the first fitness is calculated as follows: D1 is the total value of short video in the initial population, and D1 is the value of short video i itself, and f1 is the first fitness. The second fitness is calculated as follows: D2 is the total value of short video viewing effectiveness prediction in the initial population, Y j is the short video viewing effectiveness prediction value when the user browses the short video j, and f2 is the second fitness. S532, within the first preset number of iterations, according to the first fitness corresponding to the population, select the population with the first fitness greater than or equal to the first fitness threshold value, and select the first five populations with the second fitness from the population removed at this time, and save them to the backup library; When the first preset number of iterations is exceeded, the population at this time is centralized with the population in the backup library to form a new population, and the population with the second fitness greater than or equal to the preset second fitness threshold value corresponding to the population at this time is selected; S533, the selected population is obtained by the cross and mutation of the genetic algorithm; S534, after obtaining the offspring population, continue to execute S531 until the preset iteration amount is met; S535, output the offspring population as the optimal solution set of multi-objective optimization. 4.The method of claim 1, wherein: Further comprising the following steps: S6, according to the account information corresponding to each user and the corresponding second short video type set, the fan portrait corresponding to each short video type is counted, and the fan portrait includes the audience group corresponding to the short video type, the viewing time distribution information, and the distribution of the audience group on each short video platform; S7, according to the fan portrait counted, generate the short video operation strategy corresponding to each short video type, and push it to the short video operator corresponding to each short video type, and the short video operator makes the short video of the short video type according to the short video operation strategy. 5.The method of claim 4, wherein: S7 includes the following steps: S70, according to the audience group corresponding to the short video type and the distribution of the audience group on each short video platform, determine the video operator corresponding to the short video type in the short video platform with the most audience group, and obtain the first comment information of the audience group in all short videos corresponding to the video operator, and the second comment information of the non-audience group in all short videos corresponding to the video operator in the short video platform with the least audience group; S71, according to the first comment information and the second comment information, the first comment information and the second comment information are identified and screened, and it is judged whether the first comment information and the second comment information are suggestion information, if yes, it is retained, otherwise it is removed; S72, according to the current retained first comment information and second comment information, and the historical comment information retrieved from the database, determine the modification report of the short video type corresponding to the video operator, and form the corresponding short video operation strategy; S73, after forming the short video operation strategy, it is pushed to the short video operator at the first time, and the short video operator makes the short video of the short video type according to the short video operation strategy.

Images