Interactive data processing method and device for video works, storage medium, and terminal

Abstract

The present invention discloses a method and device for processing interactive data of video works, a storage medium, and a terminal, which relates to the field of media communication technology, and the main purpose is to solve the problem of low accuracy of interactive data of video works. It mainly includes obtaining the initial interactive data of the target video work published by the target account and the account category of the target account; calculating the initial interactive combination parameters of the target video work based on the initial interactive data, and determining the target distribution adjustment coefficient and the target interactive combination parameter threshold that match the account category; if an abnormal video work is identified from the target video work based on the target interactive combination parameter threshold and the initial interactive combination parameter, then calculating the abnormal rate of the interactive data of the abnormal video work based on the target distribution adjustment coefficient and the initial interactive combination parameter of the abnormal video work; correcting the initial interactive data based on the abnormal rate of the interactive data to obtain the target interactive data of the target account.

Description

Technical Field

[0001] The present invention relates to the field of media communication technology, and in particular to a method and device, a storage medium, and a terminal for interactive data processing of video works. Background Art

[0002] Short videos have become the main traffic gathering place on social media platforms. The interactive data of short videos can reflect the promotion ability of the corresponding influencers. Therefore, brand advertisers usually need to refer to the interactive data performance of the influencers' videos before cooperating with them. However, the interactive data of short videos is often artificially brushed, which makes the interactive data directly captured on the platform "watered" and cannot truly reflect the promotion ability of the influencers.

[0003] At present, for the interactive data directly captured on social media platforms, we mainly rely on experience to judge whether there is water-brushing in the interactive data. However, there is no way to confirm the extent of human intervention in the interactive data, nor can we confirm the distribution of real interactive data and water-brushing interactive data. As a result, the extracted interactive data of video works is of low accuracy, and cannot provide a reliable data basis for the analysis of influencer promotion capabilities. Summary of the invention

[0004] In view of this, the present invention provides a method and device for processing interactive data of video works, a storage medium, and a terminal, the main purpose of which is to solve the problem of low accuracy of interactive data of existing video works.

[0005] According to one aspect of the present invention, there is provided a method for processing interactive data of a video work, comprising:

[0006] Obtaining initial interaction data of a target video work published by a target account and the account category of the target account;

[0007] Calculating the initial interaction combination parameters of the target video work according to the initial interaction data, and determining the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category;

[0008] If an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter, then the interaction data abnormality rate of the abnormal video work is calculated according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work;

[0009] The initial interaction data is corrected according to the interaction data abnormality rate to obtain target interaction data of the target account.

[0010] Furthermore, before determining the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category, the method further includes:

[0011] Acquire first interaction data of video works published by multiple accounts under different account categories within a preset time period, wherein the account categories include multiple account levels in various classification fields under different platform attributes;

[0012] For each of the account categories, abnormal accounts are eliminated according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories;

[0013] For each of the account categories, an interaction combination parameter normal distribution model is constructed based on the second interaction data of the video works released by the sample accounts, and parameters are extracted based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories.

[0014] Furthermore, for each of the account categories, abnormal accounts are eliminated according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories, including:

[0015] For each of the account categories, construct a time series sample of the first interactive data, and use a time series model to perform forecasting processing on the time series sample to obtain a forecast time series, wherein the time series model is constructed based on a moving average model and an autoregressive model;

[0016] A residual sequence is calculated based on the predicted time series and the time series samples, and the first interaction data corresponding to the residual sequence that meets the sequence fluctuation amplitude threshold is determined as abnormal interaction data;

[0017] The account to which the abnormal interaction data belongs is determined as an abnormal account, and the non-abnormal accounts under the current account category are determined as sample accounts of the current account category.

[0018] Furthermore, for each of the account categories, an interaction combination parameter normal distribution model is constructed based on the second interaction data of the video works released by the sample account, and parameters are extracted based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories, including:

[0019] For each of the account categories, sample video works are extracted from the video works according to the heat decay time, and sample work interaction combination parameters of the sample video works are calculated;

[0020] Performing data fitting on the interaction combination parameters of the global sample works to obtain a normal distribution model of the interaction combination parameters;

[0021] The expected value and standard deviation of the interaction combination parameter normal distribution model are extracted as distribution adjustment coefficients, and the expected distribution interval of the interaction combination parameters of the sample works is calculated based on the expected value and the standard deviation to obtain the interaction combination parameter threshold under the current account category.

[0022] Furthermore, the target distribution adjustment coefficient is determined based on a target interaction combination parameter normal distribution model matching the account category, and the abnormal rate of interaction data of the abnormal video work is calculated based on the target distribution adjustment coefficient and the abnormal interaction combination parameter of the abnormal video work, including:

[0023] A normal deviation value is calculated based on the abnormal interaction combination parameter of the abnormal video work and the target distribution adjustment coefficient, and the normal deviation value is used to characterize the degree of deviation between the abnormal interaction combination parameter and the average value of the target interaction combination parameter normal distribution model;

[0024] The area ratio of the normal deviation value in the normal distribution model of the target interaction combination parameter is calculated to obtain the interaction data abnormality rate of the abnormal video work.

[0025] Furthermore, after the initial interaction data is corrected according to the interaction data abnormality rate to obtain the target interaction data of the target account, the method further includes:

[0026] The monitoring status of the target account is determined based on at least one of the target interaction data, the number of normal video works, the number of abnormal video works and the abnormal rate of the interaction data, and the monitoring status includes any level of abnormal status or any level of normal status.

[0027] Furthermore, the initial interaction data includes the number of likes and the number of comments, and the initial interaction combination parameters include the like-to-comment ratio or the comment-to-like ratio.

[0028] According to another aspect of the present invention, there is provided an interactive data processing device for a video work, comprising:

[0029] An acquisition module, used to acquire initial interaction data of a target video work published by a target account and an account category of the target account;

[0030] A determination module, configured to calculate the initial interaction combination parameters of the target video work according to the initial interaction data, and determine a target distribution adjustment coefficient and a target interaction combination parameter threshold that match the account category;

[0031] A calculation module, configured to calculate an abnormal rate of interaction data of the abnormal video work according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work if an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter;

[0032] The processing module is used to correct the initial interaction data according to the abnormal rate of the interaction data to obtain the target interaction data of the target account.

[0033] Furthermore, the device also includes:

[0034] The acquisition module is further used to acquire first interaction data of video works published by multiple accounts under different account categories within a preset time period, wherein the account categories include multiple account levels in various classification fields under different platform affiliations;

[0035] A removal module, configured to remove abnormal accounts for each of the account categories according to the time series characteristics of the first interactive data, and obtain sample accounts of different account categories;

[0036] A construction module is used to construct an interaction combination parameter normal distribution model for each account category based on the second interaction data of the video works released by the sample account, and to extract parameters based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories.

[0037] Furthermore, the rejection module includes:

[0038] A prediction processing unit, configured to construct a time series sample of the first interaction data for each of the account categories, and perform prediction processing on the time series sample using a time series model to obtain a predicted time series, wherein the time series model is constructed based on a moving average model and an autoregressive model;

[0039] A first calculation unit is used to calculate a residual sequence according to the predicted time series and the time series sample, and determine the first interaction data corresponding to the residual sequence that meets the sequence fluctuation amplitude threshold as abnormal interaction data;

[0040] The determination unit is used to determine the account to which the abnormal interaction data belongs as an abnormal account, and to determine the non-abnormal accounts under the current account category as sample accounts of the current account category.

[0041] Furthermore, the building block includes:

[0042] An extraction unit, configured to extract sample video works from the video works according to the heat decay time for each of the account categories, and calculate sample work interaction combination parameters of the sample video works;

[0043] A fitting unit, used for performing data fitting on the interaction combination parameters of the global sample works to obtain a normal distribution model of the interaction combination parameters;

[0044] The second calculation unit is used to extract the expected value and standard deviation of the normal distribution model of the interaction combination parameters as the distribution adjustment coefficient, and calculate the expected distribution interval of the interaction combination parameters of the sample works based on the expected value and the standard deviation to obtain the interaction combination parameter threshold under the current account category.

[0045] Furthermore, the calculation module includes:

[0046] A third calculation unit is used to calculate a normal deviation value according to the abnormal interaction combination parameter of the abnormal video work and the target distribution adjustment coefficient, wherein the normal deviation value is used to characterize the degree of deviation between the abnormal interaction combination parameter and the average value of the target interaction combination parameter normal distribution model;

[0047] The fourth calculation unit is used to calculate the area ratio of the normal deviation value in the normal distribution model of the target interaction combination parameter, and obtain the interaction data abnormality rate of the abnormal video work.

[0048] Furthermore, the device also includes:

[0049] The monitoring status determination module is used to determine the monitoring status of the target account based on at least one of the target interaction data, the number of normal video works, the number of abnormal video works and the abnormal rate of the interaction data, and the monitoring status includes any level of abnormal status or any level of normal status.

[0050] Furthermore, in a specific application scenario, the initial interaction data includes the number of likes and the number of comments, and the initial interaction combination parameter includes the like-to-comment ratio or the comment-to-like ratio.

[0051] According to another aspect of the present invention, there is provided a storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the interactive data processing method of the video work as described above.

[0052] According to another aspect of the present invention, there is provided a terminal, comprising: a processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus;

[0053] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the interactive data processing method of the above-mentioned video work.

[0054] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0055] The present invention provides a method and device for processing interactive data of a video work, a storage medium, and a terminal. The embodiment of the present invention obtains the initial interactive data of a target video work published by a target account and the account category of the target account; calculates the initial interactive combination parameters of the target video work based on the initial interactive data, and determines the target distribution adjustment coefficient and the target interactive combination parameter threshold that match the account category; if an abnormal video work is identified from the target video work based on the target interactive combination parameter threshold and the initial interactive combination parameters, the interactive data abnormality rate of the abnormal video work is calculated based on the target distribution adjustment coefficient and the initial interactive combination parameters of the abnormal video work; the initial interactive data is corrected based on the interactive data abnormality rate to obtain the target interactive data of the target account, which greatly reduces the impact of human intervention on the interactive data, reduces the "water content" of the interactive data, and thus greatly improves the accuracy of the interactive data.

[0056] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0058] Figure 1 A flow chart of an interactive data processing method for a video work provided by an embodiment of the present invention is shown;

[0059] Figure 2 A flow chart of another interactive data processing method for a video work provided by an embodiment of the present invention is shown;

[0060] Figure 3 A normal interactive data fluctuation diagram provided by an embodiment of the present invention is shown;

[0061] Figure 4An abnormal interactive data fluctuation diagram provided by an embodiment of the present invention is shown;

[0062] Figure 5 A block diagram showing the composition of an interactive data processing device for video works provided by an embodiment of the present invention is shown;

[0063] Figure 6 A schematic diagram of the structure of a terminal provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0064] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0065] In order to solve the problem of low accuracy of interactive data of video works, the embodiment of the present invention provides a method for processing interactive data of video works, such as Figure 1 As shown, the method includes:

[0066] 101. Obtain initial interaction data of a target video work published by a target account and an account category of the target account.

[0067] In the embodiment of the present invention, the target account is the account of the creator of the video work on the social media platform. The target video work is the video work published by the target account. The content of the video work can be events, life views, products, etc., which are not specifically limited in the embodiment of the present invention. The target video work can be a video work published by the target account within a preset time interval, or it can be all video works published by the target account. The initial interaction data is collected directly from the public data of the platform page. The interaction data corresponding to the target video work can include the number of collections, likes, reposts and comments of the video work. The account category of the target account can be divided according to the different social media platforms where the target account is located, according to the fields involved in the target account, or according to the number of fans of the target account. Of course, it can also be divided into multiple levels according to two or three of the above three aspects. For example, accounts with more than 3.5 million fans in the field of A social media platform and b are divided into one account category, and accounts with fans in the range of 1.5 million to 3.5 million in the field of A social media platform and b are divided into another account category.

[0068] 102. Calculate the initial interaction combination parameters of the target video work according to the initial interaction data, and determine the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category;

[0069] In the embodiment of the present invention, the initial interaction combination parameter is a parameter obtained by combining and calculating two or two interaction data, for example, the ratio of the number of likes to the number of comments, the ratio of the number of likes to the number of reposts, etc. The target distribution adjustment coefficient and the target interaction combination parameter threshold are the distribution adjustment coefficient and the interaction combination parameter threshold that match the account category of the target account, and can be obtained based on the mapping relationship between the account category and the distribution adjustment coefficient and the interaction combination parameter threshold. Among them, the interaction combination parameter threshold characterizes the normal distribution characteristics of the interaction combination parameters of the video works published by the account of the corresponding account category. For example, on the A social media platform, the normal distribution range of the likes and ratings of the video works of the account with more than 3.5 million fans in the field of b is 10-30, and the corresponding interaction combination parameter threshold is the first interaction combination parameter threshold of 10 and the second interaction combination parameter threshold of 30. The distribution adjustment coefficient is used to convert the initial interaction combination parameter into a deviation distance from the mean of the target normal distribution model of the interaction combination parameter, so as to determine the abnormal rate of the interaction data of the video work based on this deviation distance.

[0070] 103. If an abnormal video work is identified from the target video work based on the target interaction combination parameter threshold and the initial interaction combination parameter, the interaction data abnormality rate of the abnormal video work is calculated based on the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work.

[0071] In the embodiment of the present invention, since the number of likes is the most important indicator in the page public data of each social media platform, a higher number of likes helps to increase the exposure rate of the video work, that is, the recommendation rate, so as to obtain platform support traffic and attract more user attention. Therefore, "brushing water" generally chooses to brush "likes", resulting in a significant increase in the number of likes. For video works for the same category of video accounts, their various interactive data change synchronously with a relatively stable ratio. For example, as the number of likes increases, the number of comments will also increase synchronously at a certain ratio. Therefore, the initial interactive data and the abnormality of the target video work can be judged by judging whether the initial interactive combination parameters meet the target interactive combination parameter threshold. The target video work can be all normal video works, all abnormal video works, and also include some normal video works and some abnormal video works. If all are normal works, there is no need to correct the initial interactive data. The abnormal rate of interactive data is used to characterize the proportion of abnormal interactive data of each abnormal video work to the initial interactive data.

[0072] It should be noted that due to the different patterns of interaction data changes presented by accounts on different social media platforms, in different industries and with different numbers of fans, the thresholds used to determine the normality and abnormality of interaction data, as well as the adjustment coefficients used to determine the abnormality rate of interaction data are also different. Therefore, it is necessary to pre-classify the account categories of video accounts, and extract the corresponding distribution adjustment coefficients and interaction combination parameter thresholds based on the interaction data of several accounts under different account categories. This can improve the accuracy of distinguishing normal interaction data from abnormal interaction data and the accuracy of calculating the abnormality rate of interaction data.

[0073] 104. Correct the initial interaction data according to the interaction data abnormality rate to obtain target interaction data of the target account.

[0074] In an embodiment of the present invention, when there are abnormal video works in the target video works, it is necessary to calculate the abnormal interaction data abnormality rate of the corresponding abnormal video works based on the abnormal interaction data of the abnormal video works, and correct the initial interaction data based on the abnormal interaction data rate. For example, by calculating the product of the abnormal interaction data rate and the initial number of likes, the abnormal number of likes is obtained, and then the difference between the initial number of likes and the abnormal number of likes is calculated to obtain the abnormal interaction data filtering result of the corresponding abnormal video works, that is, the corrected initial interaction data. The target interaction data may include the corrected initial interaction data of each abnormal video work and / or the initial interaction data of the normal video work, and may also include the cumulative result of the above two types of interaction data.

[0075] In an embodiment of the present invention, for further explanation and limitation, as Figure 2 As shown, the step determines the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category. Before that, the method also includes:

[0076] 201. Obtaining first interaction data of video works published by multiple accounts under different account categories within a preset time period;

[0077] 202. For each of the account categories, remove abnormal accounts according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories;

[0078] 203. For each of the account categories, a normal distribution model of interaction combination parameters is constructed based on the second interaction data of the video works released by the sample accounts, and parameters are extracted based on the normal distribution model of interaction combination parameters to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories.

[0079] In an embodiment of the present invention, the account category includes multiple account levels in various classification fields under different platform affiliations. Due to the different ecosystems of different social platforms and the differences in the magnitude of interactive data, it is necessary to divide the social platforms where the accounts are located. Specifically, a social platform can be divided into a category, for example, a certain audio, a certain book, etc. The fields involved in the account may include beauty, fashion, drama, automobiles, technology, finance, etc. The account level can be divided according to the level of fans of the account. For example, the head level is more than 3.5 million fans, the shoulder level is between 1.5 million and 3.5 million fans, the waist level is between 600,000 and 1.5 million fans, and the tail level is less than 600,000 fans. In order to extract the characteristics of interactive data under different account categories, for each account category, the first interactive data of video works released by multiple (such as 500) accounts are extracted according to the preset time period, so as to perform data feature analysis based on the first interactive data to determine the interactive combination parameter threshold and distribution adjustment coefficient of the current account category. Since the first interactive data is actually collected by the social platform, it is possible that the interactive data of the account may be "watered". Therefore, it is necessary to filter abnormal accounts and filter out the interactive data of abnormal accounts from the first interactive data to provide accurate interactive data basis for subsequent feature analysis. The preset time period can be 6 months, or it can be customized according to actual application requirements, which is not specifically limited in the embodiment of the present invention.

[0080] It should be noted that, through the chart method, the fluctuation trend charts of the number of likes of the video works released by the marked normal accounts and the marked abnormal accounts at different times are drawn, such as Figure 3 , 4 As shown in the figure, the fluctuation of the number of likes of normal accounts does not show seasonal or cyclical trends, and the fluctuation trend of the number of likes is non-stationary, while the fluctuation of the number of likes of abnormal accounts has a cyclical law over time, and the number of likes fluctuates steadily in the time dimension. Through the statistical judgment method, the number of likes of normal accounts and abnormal accounts is analyzed week-on-week or month-on-month, and month-on-month respectively. It is found that the number of likes of normal accounts also shows a non-stationary fluctuation trend, while the number of likes of abnormal accounts shows a stable fluctuation trend. Combining the analysis conclusions of the two methods, for normal accounts, the number of likes of works changes over time, which does not have a cyclical law, and the change of account interaction data over time is non-stationary fluctuation. Therefore, based on the time series characteristics of the interaction data corresponding to the account, the normal and abnormal accounts in the account can be detected, and the normal and abnormal accounts in the account can be accurately identified, and the abnormal accounts can be accurately filtered.

[0081] In one embodiment of the present invention, for further explanation and limitation, the step of performing abnormal account elimination processing for each account category according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories includes:

[0082] For each of the account categories, construct a time series sample of the first interactive data, and use a time series model to perform prediction processing on the time series sample to obtain a predicted time series;

[0083] A residual sequence is calculated based on the predicted time series and the time series samples, and the first interaction data corresponding to the residual sequence that meets the sequence fluctuation amplitude threshold is determined as abnormal interaction data;

[0084] The account to which the abnormal interaction data belongs is determined as an abnormal account, and the non-abnormal accounts under the current account category are determined as sample accounts of the current account category.

[0085] In an embodiment of the present invention, a time series model is constructed based on a moving average model and an autoregressive model. Specifically, a 6-month time series is modeled using a moving average model and an autoregressive model to obtain a time series model. The time series sample is input into the time series model, and the predicted value at each time point is calculated, that is, the predicted time series, and then the residuals of the two sequences at each time point are calculated to obtain a residual sequence. After obtaining the residual sequence, a residual sequence test is performed, that is, whether the residual sequence satisfies the expectation of the residual being 0, the variance of the residual being stable, and the residual having no autocorrelation. If all of them are satisfied, it indicates that the residual sequence of the current account satisfies the stationarity, and the time series sample is also stationary, then the current account is an abnormal account. Otherwise, the current account is a normal account.

[0086] In one embodiment of the present invention, for further explanation and limitation, the steps are as follows: for each of the account categories, a normal distribution model of interaction combination parameters is constructed based on the second interaction data of the video works published by the sample account, and parameters are extracted based on the normal distribution model of interaction combination parameters to obtain interaction combination parameter thresholds and distribution adjustment coefficients of different account categories, including:

[0087] For each of the account categories, sample video works are extracted from the video works according to the heat decay time, and sample work interaction combination parameters of the sample video works are calculated;

[0088] Performing data fitting on the interaction combination parameters of the global sample works to obtain a normal distribution model of the interaction combination parameters;

[0089] The expected value and standard deviation of the interaction combination parameter normal distribution model are extracted as distribution adjustment coefficients, and the expected distribution interval of the interaction combination parameters of the sample works is calculated based on the expected value and the standard deviation to obtain the interaction combination parameter threshold under the current account category.

[0090] In the embodiment of the present invention, the heat decay time is used to characterize the life decay time of the video work, that is, the video work will generate less interactive data after the heat decay time. According to the research on the heat of the video work, the heat decay time can be configured as 21 natural days, or it can be customized according to the heat decay situation of the actual video work, which is not specifically limited in the embodiment of the present invention. In order to ensure the integrity and stability of the interactive data of the video work, before constructing the interactive combination parameter normal distribution model based on the second interactive data of the video work published by the sample account, it is necessary to screen the video works, and take the video works whose publishing time interval has exceeded the heat decay time, that is, the interactive data has been collected completely as sample video works, and calculate the interactive combination parameters of the sample works based on the second interactive data of the sample video work. The construction of the interactive combination parameter normal distribution model can be completed based on the normal distribution fitting tool, such as the SciPy library in Python. After obtaining the expected value and standard deviation of the normal distribution model, according to the interactive combination parameter distribution characteristics characterized by the expected value and standard deviation and Chebyshev's theorem, the distribution area of ​​89% of the interactive combination parameters can be determined, and then the critical values ​​at both ends of this distribution area can be determined as the interactive combination parameter threshold. For example, if the expected value is μ=16.7 and the standard deviation is σ=4.87, then 89% of the interaction combination parameters are distributed in the range of μ±σ, i.e. (11.83, 21.57). The minimum value of the interaction combination parameter threshold is 11.83 and the maximum value is 21.57. The statistical data of the likes and comments ratios of 120,002 video works are shown in Table 1. Except for a small number of data outliers that are too large or too small, the data distribution is mainly concentrated between 10-40, which conforms to the typical normal distribution characteristics. Therefore, the individual outliers in the interaction combination parameters can be removed to construct a normal distribution model.

[0091] Table 1:

[0092]

[0093]

[0094] It should be noted that the expected value is the maximum likelihood estimate calculated based on the sample data, which can be understood as the average level of the sample and represents the best estimate of the overall "true level". By determining the interaction combination parameter threshold based on the expected value, the normal interaction data video works and abnormal interaction data video works can be better distinguished, thereby improving the accuracy of recognition.

[0095] In one embodiment of the present invention, for further explanation and limitation, the step of calculating the abnormal rate of interaction data of the abnormal video work according to the target distribution adjustment coefficient and the abnormal interaction combination parameter of the abnormal video work includes:

[0096] A normal deviation value is calculated based on the abnormal interaction combination parameter of the abnormal video work and the target distribution adjustment coefficient;

[0097] The area ratio of the normal deviation value in the normal distribution model of the target interaction combination parameter is calculated to obtain the interaction data abnormality rate of the abnormal video work.

[0098] In the embodiment of the present invention, the target distribution adjustment coefficient is determined based on the target interaction combination parameter normal distribution model that matches the account category. The normal deviation value is used to characterize the degree of deviation of the abnormal interaction combination parameter from the average value of the target interaction combination parameter normal distribution model. The normal deviation value is set to z, the target distribution adjustment coefficient is the expected value μ1 and σ1 of the target interaction combination parameter normal distribution model, and the calculation formula of the normal deviation value is:

[0099]

[0100] Among them, x is the interaction combination parameter of each video work. After determining the normal deviation value, the area ratio of the current normal deviation value in the normal distribution model of the target interaction combination parameter can be determined by the scipy.stats.norm.cdf function of the scipy library in Python or by looking up a table. This area ratio is the abnormal rate of the interaction data of the abnormal video work.

[0101] In an application example, the expert account to be detected has released 4 video works within 6 months. The interaction combination parameter threshold corresponding to the account category is (11.83, 21.57), the distribution adjustment coefficient is μ = 16.7, σ = 4.87, and the normal deviation value is set to Z. The interaction data is shown in Table 2 below:

[0102] Table 2:

[0103]

[0104] Judging whether the work is watered based on the likes and ratings: For work 2 and work 3, the likes and ratings are within (11.83, 21.57), so there is an 89% probability that the work is watered. Works 1 and 4 are highly suspected of being watered, and the water content is tested. Using the normal distribution parameters of μ=16.7, σ=4.87, the conversion is: Z of work 1 is (25-16.7) / 4.87=1.71; Z of work 2 is 0.68; Z of work 3 is 0.68; Z of work 4 is 23.22. Work 1: 500 likes, Z score 1.71, the corresponding area share (can be checked in the table) is 4.4%. Therefore, the number of watered likes is 500x 4.4%≈22, and the water content (interaction data abnormality rate) = 22 / 500≈4.4%. The water content of work 3 is close to 100%.

[0105] In one embodiment of the present invention, for further explanation and limitation, after the step of correcting the initial interaction data according to the abnormal rate of the interaction data to obtain the target interaction data of the target account, the method further includes:

[0106] The monitoring status of the target account is determined based on at least one of the target interaction data, the number of normal video works, the number of abnormal video works, and the abnormal rate of the interaction data.

[0107] In the embodiment of the present invention, the monitoring state includes any level of abnormal state or any level of normal state. For example, if the ratio of the number of abnormal video works to the number of normal video works is less than 50% and greater than 30%, the monitoring state is determined to be level three abnormal; if it is greater than 50% and less than 70%, the monitoring state is determined to be level two abnormal; if it is greater than 70%, the monitoring state is determined to be level one abnormal. For another example, if the number of abnormal video works is zero, the monitoring state is determined to be level one normal, and if the ratio of the number of abnormal video works to the number of normal video works is less than 10%, the monitoring state is determined to be level two normal. Similarly, the monitoring state level can also be determined according to the abnormal rate of interactive data. In addition, after determining a level of monitoring state based on the abnormal rate of interactive data, the level of the monitoring state can be adjusted based on the size of the target interactive data volume. For example, the monitoring state determined based on the abnormal rate of interactive data is level two normal, but the target interactive data volume exceeds the preset interactive volume threshold (such as 1 million). Considering that the actual interactive data volume is also large, the monitoring state can be corrected to level one normal. Of course, it can also be downgraded. The specific adjustment method can be adjusted according to the actual application requirements, and the embodiment of the present invention is not specifically limited.

[0108] In one embodiment of the present invention, for further explanation and limitation, the interaction data includes at least one of the number of likes, the number of comments and the number of reposts, and the initial interaction combination parameter includes the like-to-comment ratio or the comment-to-like ratio.

[0109] In the embodiment of the present invention, the like-to-comment ratio is preferably used as the interaction combination parameter, which can more accurately reflect the "water content" of the interaction data. Of course, the like-to-comment ratio or the like-to-repost ratio can also be used as the interaction combination parameter to execute the above method. The preferred interaction data is the number of likes and the number of comments. Any combination of the number of likes, the number of comments, and the number of reposts can also be selected, which is not specifically limited in the embodiment of the present invention.

[0110] The present invention provides a method for processing interactive data of a video work. An embodiment of the present invention obtains initial interactive data of a target video work published by a target account and an account category of the target account; calculates initial interactive combination parameters of the target video work based on the initial interactive data, and determines a target distribution adjustment coefficient and a target interactive combination parameter threshold that match the account category; if an abnormal video work is identified from the target video work based on the target interactive combination parameter threshold and the initial interactive combination parameters, calculates the interactive data anomaly rate of the abnormal video work based on the target distribution adjustment coefficient and the initial interactive combination parameters of the abnormal video work; corrects the initial interactive data based on the interactive data anomaly rate to obtain the target interactive data of the target account, thereby greatly reducing the impact of human intervention on the interactive data and reducing the "water content" of the interactive data, thereby greatly improving the accuracy of the interactive data.

[0111] Furthermore, as a response to the above Figure 1 The embodiment of the present invention provides an interactive data processing device for video works, such as Figure 5 As shown, the device comprises:

[0112] An acquisition module 31 is used to acquire initial interaction data of a target video work published by a target account and an account category of the target account;

[0113] A determination module 32 is used to calculate the initial interaction combination parameters of the target video work according to the initial interaction data, and determine the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category;

[0114] A calculation module 33, configured to calculate an abnormal interaction data rate of the abnormal video work according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work if an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter;

[0115] The processing module 34 is used to correct the initial interaction data according to the interaction data abnormality rate to obtain the target interaction data of the target account.

[0116] Furthermore, the device also includes:

[0117] The acquisition module is further used to acquire first interaction data of video works published by multiple accounts under different account categories within a preset time period, wherein the account categories include multiple account levels in various classification fields under different platform affiliations;

[0118] A removal module, configured to remove abnormal accounts for each of the account categories according to the time series characteristics of the first interactive data, and obtain sample accounts of different account categories;

[0119] A construction module is used to construct an interaction combination parameter normal distribution model for each account category based on the second interaction data of the video works released by the sample account, and to extract parameters based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories.

[0120] Furthermore, the rejection module includes:

[0121] A prediction processing unit, configured to construct a time series sample of the first interaction data for each of the account categories, and perform prediction processing on the time series sample using a time series model to obtain a predicted time series, wherein the time series model is constructed based on a moving average model and an autoregressive model;

[0122] A first calculation unit is used to calculate a residual sequence according to the predicted time series and the time series sample, and determine the first interaction data corresponding to the residual sequence that meets the sequence fluctuation amplitude threshold as abnormal interaction data;

[0123] The determination unit is used to determine the account to which the abnormal interaction data belongs as an abnormal account, and to determine the non-abnormal accounts under the current account category as sample accounts of the current account category.

[0124] Furthermore, the building block includes:

[0125] An extraction unit, configured to extract sample video works from the video works according to the heat decay time for each of the account categories, and calculate sample work interaction combination parameters of the sample video works;

[0126] A fitting unit, used for performing data fitting on the interaction combination parameters of the global sample works to obtain a normal distribution model of the interaction combination parameters;

[0127] The second calculation unit is used to extract the expected value and standard deviation of the normal distribution model of the interaction combination parameters as the distribution adjustment coefficient, and calculate the expected distribution interval of the interaction combination parameters of the sample works based on the expected value and the standard deviation to obtain the interaction combination parameter threshold under the current account category.

[0128] Furthermore, the calculation module 33 includes:

[0129] A third calculation unit is used to calculate a normal deviation value according to the abnormal interaction combination parameter of the abnormal video work and the target distribution adjustment coefficient, wherein the normal deviation value is used to characterize the degree of deviation between the abnormal interaction combination parameter and the average value of the target interaction combination parameter normal distribution model;

[0130] The fourth calculation unit is used to calculate the area ratio of the normal deviation value in the normal distribution model of the target interaction combination parameter, and obtain the interaction data abnormality rate of the abnormal video work.

[0131] Furthermore, the device also includes:

[0132] The monitoring status determination module is used to determine the monitoring status of the target account based on at least one of the target interaction data, the number of normal video works, the number of abnormal video works and the abnormal rate of the interaction data, and the monitoring status includes any level of abnormal status or any level of normal status.

[0133] Furthermore, in a specific application scenario, the initial interaction data includes the number of likes and the number of comments, and the initial interaction combination parameter includes the like-to-comment ratio or the comment-to-like ratio.

[0134] The present invention provides an interactive data processing device for a video work. An embodiment of the present invention obtains initial interactive data of a target video work published by a target account and an account category of the target account; calculates initial interactive combination parameters of the target video work based on the initial interactive data, and determines a target distribution adjustment coefficient and a target interactive combination parameter threshold that match the account category; if an abnormal video work is identified from the target video work based on the target interactive combination parameter threshold and the initial interactive combination parameters, calculates the interactive data anomaly rate of the abnormal video work based on the target distribution adjustment coefficient and the initial interactive combination parameters of the abnormal video work; corrects the initial interactive data based on the interactive data anomaly rate to obtain the target interactive data of the target account, thereby greatly reducing the impact of human intervention on the interactive data and reducing the "water content" of the interactive data, thereby greatly improving the accuracy of the interactive data.

[0135] According to an embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the interactive data processing method of the video work in any of the above method embodiments.

[0136] Figure 6 A schematic diagram of the structure of a terminal provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the terminal.

[0137] like Figure 6As shown, the terminal may include: a processor (processor) 402 , a communication interface (Communications Interface) 404 , a memory (memory) 406 , and a communication bus 408 .

[0138] The processor 402 , the communication interface 404 , and the memory 406 communicate with each other via a communication bus 408 .

[0139] The communication interface 404 is used to communicate with other devices such as clients or other servers.

[0140] The processor 402 is used to execute the program 410, and specifically can execute the relevant steps in the embodiment of the interactive data processing method of the above-mentioned video work.

[0141] Specifically, the program 410 may include program codes, which include computer operation instructions.

[0142] The processor 402 may be a central processing unit (CPU), or an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiment of the present invention. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs; or processors of different types, such as one or more CPUs and one or more ASICs.

[0143] The memory 406 is used to store the program 410. The memory 406 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0144] The program 410 may be specifically configured to enable the processor 402 to perform the following operations:

[0145] Obtaining initial interaction data of a target video work published by a target account and the account category of the target account;

[0146] Calculating the initial interaction combination parameters of the target video work according to the initial interaction data, and determining the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category;

[0147] If an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter, then the interaction data abnormality rate of the abnormal video work is calculated according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work;

[0148] The initial interaction data is corrected according to the interaction data abnormality rate to obtain target interaction data of the target account.

[0149] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0150] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for processing interactive data of a video work, characterized in that: include: Obtaining initial interaction data of a target video work published by a target account and the account category of the target account; Calculate the initial interaction combination parameters of the target video work according to the initial interaction data, and determine the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category, wherein the target distribution adjustment coefficient is determined based on a normal distribution model of the target interaction combination parameters that match the account category; If an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter, then the interaction data abnormality rate of the abnormal video work is calculated according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work; The initial interaction data is corrected according to the interaction data abnormality rate to obtain target interaction data of the target account.

2. The method according to claim 1, characterized in that Before determining the target distribution adjustment coefficient and target interaction combination parameter threshold that match the account category, the method further includes: Acquire first interaction data of video works published by multiple accounts under different account categories within a preset time period, wherein the account categories include multiple account levels in various classification fields under different platform attributes; For each of the account categories, abnormal accounts are eliminated according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories; For each of the account categories, an interaction combination parameter normal distribution model is constructed based on the second interaction data of the video works released by the sample accounts, and parameters are extracted based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients for different account categories.

3. The method according to claim 2, characterized in that For each of the account categories, abnormal accounts are eliminated according to the time series characteristics of the first interactive data to obtain sample accounts of different account categories, including: For each of the account categories, construct a time series sample of the first interactive data, and use a time series model to perform forecasting processing on the time series sample to obtain a forecast time series, wherein the time series model is constructed based on a moving average model and an autoregressive model; A residual sequence is calculated based on the predicted time series and the time series samples, and the first interaction data corresponding to the residual sequence that meets the sequence fluctuation amplitude threshold is determined as abnormal interaction data; The account to which the abnormal interaction data belongs is determined as an abnormal account, and the non-abnormal accounts under the current account category are determined as sample accounts of the current account category.

4. The method according to claim 2, characterized in that: For each of the account categories, an interaction combination parameter normal distribution model is constructed based on the second interaction data of the video works released by the sample account, and parameters are extracted based on the interaction combination parameter normal distribution model to obtain interaction combination parameter thresholds and distribution adjustment coefficients of different account categories, including: For each of the account categories, sample video works are extracted from the video works according to the heat decay time, and sample work interaction combination parameters of the sample video works are calculated; Performing data fitting on the interaction combination parameters of the global sample works to obtain a normal distribution model of the interaction combination parameters; The expected value and standard deviation of the interaction combination parameter normal distribution model are extracted as distribution adjustment coefficients, and the expected distribution interval of the interaction combination parameters of the sample works is calculated based on the expected value and the standard deviation to obtain the interaction combination parameter threshold under the current account category.

5. The method according to claim 1, characterized in that The abnormal rate of interaction data of the abnormal video work is calculated based on the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work, including: A normal deviation value is calculated based on the abnormal interaction combination parameter of the abnormal video work and the target distribution adjustment coefficient, and the normal deviation value is used to characterize the degree of deviation between the abnormal interaction combination parameter and the average value of the target interaction combination parameter normal distribution model; The area ratio of the normal deviation value in the normal distribution model of the target interaction combination parameter is calculated to obtain the interaction data abnormality rate of the abnormal video work.

6. The method according to claim 1, characterized in that After the initial interaction data is corrected according to the interaction data abnormality rate to obtain the target interaction data of the target account, the method further includes: The monitoring status of the target account is determined based on at least one of the target interaction data, the number of normal video works, the number of abnormal video works and the abnormal rate of the interaction data, and the monitoring status includes any level of abnormal status or any level of normal status.

7. The method according to any one of claims 1 to 6, characterized in that The initial interaction data includes the number of likes and the number of comments, and the initial interaction combination parameters include the like-to-comment ratio or the comment-to-like ratio.

8. An interactive data processing device for video works, characterized in that: include: An acquisition module, used to acquire initial interaction data of a target video work published by a target account and an account category of the target account; A determination module, configured to calculate the initial interaction combination parameters of the target video work according to the initial interaction data, and determine a target distribution adjustment coefficient and a target interaction combination parameter threshold that match the account category, wherein the target distribution adjustment coefficient is determined based on a normal distribution model of the target interaction combination parameters that match the account category; A calculation module, configured to calculate an abnormal rate of interaction data of the abnormal video work according to the target distribution adjustment coefficient and the initial interaction combination parameter of the abnormal video work if an abnormal video work is identified from the target video work according to the target interaction combination parameter threshold and the initial interaction combination parameter; The processing module is used to correct the initial interaction data according to the abnormal rate of the interaction data to obtain the target interaction data of the target account.

9. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the interactive data processing method for a video work as described in any one of claims 1-7.

10. A terminal, comprising: A processor, a memory, a communication interface and a communication bus, wherein the processor, the memory and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute an operation corresponding to the interactive data processing method of a video work as described in any one of claims 1-7.

Images