Network abnormal behavior detection system based on big data
Through a network abnormal behavior detection system based on big data, combined with playback volume and user behavior analysis, it can identify and identify malicious playback behavior in video websites, and solve the problem of malicious playback volume in video websites, improving detection accuracy and prevention Effect.
Patent Information
- Application Number
- CN202210414622.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2020-12-02
- Publication Date
- 2024-11-01
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
There is a problem of malicious playback volume in video websites, which seriously damages the interests of video website platforms and video producers. It is difficult for the existing technology to effectively detect and prevent such abnormal behavior.
A network abnormal behavior detection system based on big data is designed, and a judgment module is added through the first suspected abnormal video judgment module and the pre-identification. Combining the playback volume reference value, abnormal index and user behavior, it determines whether the video file and user have abnormal behavior. and give corresponding marks.
The accuracy of detection of abnormal video playback volume is improved, not only considering the video playback volume, but also analyzing the historical playback status of the video uploader to prevent one-sided evaluation and ensuring effective identification and identification of malicious playback behavior.
Smart Images

Figure CN114979778B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of big data, and in particular to a network abnormal behavior detection system based on big data. Background Art
[0002] With the rapid development of the Internet, video websites have gradually become familiar and used by the public. As video websites become more open to users, many video producers will upload some videos they shoot on a daily basis to video websites to share with everyone. However, in order to increase popularity or improve economic benefits, there are often behaviors on the Internet where a certain video file is repeatedly played on demand. This malicious playback volume is considered abnormal playback volume, which seriously damages the interests of the video website platform and other video producers. Therefore, detecting whether a video playback is abnormal playback is an urgent problem to be solved. Summary of the invention
[0003] The purpose of the present invention is to provide a network abnormal behavior detection system and method based on big data to solve the problems raised in the above background technology.
[0004] In order to solve the above technical problems, the present invention provides the following technical solutions: a network abnormal behavior detection system based on big data, the detection system comprising a first suspected abnormal video judgment module and a pre-identification adding judgment module, the first suspected abnormal video judgment module obtains the time interval between the current time and the time when the video file is uploaded when the user detects that the playback volume of a certain video meets certain conditions, and when the time interval is less than or equal to the time interval threshold, the video file is judged to be a first suspected abnormal video, and the pre-identification adding judgment module collects the playback status of historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judges whether to add an abnormal pre-identification to the first suspected abnormal video based on this.
[0005] Furthermore, the first suspected abnormal video judgment module includes a playback volume reference value calculation module and a playback volume reference value comparison module. The playback volume reference value calculation module calculates a playback volume reference value Hc=Hf / Hz according to the total playback volume Hz of the video file and the playback volume Hf of the video file in a preset time period. The playback volume reference value comparison module compares the relationship between the playback volume reference value and the playback volume reference threshold value. When the playback volume reference value is greater than or equal to the playback volume reference threshold value, the time interval between the current time and the time when the video file is uploaded is obtained; the pre-identification adding judgment module includes a reference video selection module, a first abnormal index calculation module, a second abnormal index calculation module and a second abnormal index comparison module. The reference video selection module collects the playback status of the historical videos of the uploader of the first suspected abnormal video, and compares the historical videos. The historical videos are sorted in descending order according to the total amount of playback, and the top three videos are selected in turn as the first reference video, the second reference video and the third reference video of the first suspected abnormal video. The first abnormal index calculation module collects the playback conditions of the first suspected abnormal video, the first reference video, the second reference video and the third reference video, and calculates the first abnormal index of each video according to the playback conditions. The second abnormal index calculation module calculates the second abnormal index of the first suspected abnormal video according to the abnormal indexes of the first suspected abnormal video, the first reference video, the second reference video and the third reference video. The second abnormal index comparison module is used to determine the relationship between the second abnormal index and the second abnormal index threshold. When the second abnormal index is greater than or equal to the second abnormal index threshold, a video abnormality pre-marker is added to the playback video.
[0006] Furthermore, the first abnormal index calculation module includes a video playback statistics module and a first abnormal index output module, the video playback statistics module is used to respectively count the total number of playbacks Pz, the first doubtful number of playbacks P1, the second doubtful number of playbacks P2, the third doubtful number of playbacks P3, the total number of users Rz who play the video, and the number of users Ry with abnormal user identifiers added of a certain video, and the first abnormal index output module calculates the first abnormal index of the video according to the statistical results of the video playback statistics module, wherein, when the video is played at a certain time, the ratio of the playback time of the video to the total duration of the video is less than or equal to the first ratio threshold, the video played at this time is the first doubtful playback, when the video is played at a certain time, the video is played at the maximum speed, the video played at this time is the second doubtful playback, and the link to the upper-level page of the video played at a certain time, if the upper-level page link belongs to a website other than the video upload website, the video played at this time is the third doubtful playback.
[0007] Furthermore, the detection system also includes an abnormal user identification adding module and an abnormal upload identification adding module, the abnormal user identification adding module includes an abnormal parameter calculation module and an abnormal parameter comparison module, the abnormal parameter calculation module is used to obtain the total number of historical video playback times Yz of a certain user, the number of times the user plays videos at the maximum speed Yb, and the number of times when a certain video is played, the ratio of the playing time of the video to the total time of the video is less than or equal to the second ratio threshold value, and calculate the abnormal parameters of the user based on this, the abnormal parameter comparison module is used to compare the abnormal parameters with the abnormal parameter threshold value, when the abnormal parameters are greater than or equal to the abnormal parameter threshold value, then add an abnormal user identification to the user, the abnormal upload identification adding module includes a pre-identification ratio calculation module and a pre-identification ratio comparison module, the pre-identification ratio calculation module counts the number of all videos Gz of the uploader of the first suspected abnormal video and the number of videos with abnormal pre-identification Gy, and calculates the pre-identification ratio Gb=Gy / Gz based on this, the pre-identification ratio comparison module compares the pre-identification ratio with the pre-identification ratio threshold value, and when the pre-identification ratio is greater than or equal to the pre-identification ratio threshold value, adds an abnormal upload identification to the uploader.
[0008] A method for detecting abnormal network behavior based on big data, the detection method comprising the following steps:
[0009] Step S1: When the playback volume of a certain video meets certain conditions, the time interval between the current time and the time when the video file is uploaded is obtained. If the time interval is less than or equal to the time interval threshold, the video file is the first suspected abnormal video.
[0010] Step S2: Collect the playback status of historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judge whether to add an abnormal pre-marker to the first suspected abnormal video based on the information.
[0011] Furthermore, when the playback volume of a certain video meets certain conditions in step S1, the following steps are included:
[0012] Get the total playback volume Hz of a certain video file and the playback volume Hf of the video file in a preset time period, calculate the playback volume reference value Hc=Hf / Hz, and when the playback volume reference value is greater than or equal to the playback volume reference threshold, get the time interval between the current time and the time when the video file was uploaded. When preliminarily judging whether the playback volume of a certain video file may be abnormal, by considering both the playback volume reference value and the time interval, the technical effect of improving the accuracy of judging whether the video file may be abnormal is achieved;
[0013] Furthermore, the step S2 further includes the following:
[0014] Step S21: Collect the playback status of historical videos of the uploader of the first suspected abnormal video, sort the historical videos in descending order according to the total amount of playback, and select the top three videos in turn as the first reference video, the second reference video and the third reference video of the first suspected abnormal video;
[0015] Step S22: collecting playback conditions of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video, and calculating the first abnormality index of each video according to the playback conditions;
[0016] Step S23: the second abnormality index of the first suspected abnormal video
[0017] Wv=0.72*W1a+0.28*(0.5*W1b+0.3*W1c+0.2*W1d),
[0018] Wherein, W1a, W1b, W1c, and W1d are the abnormality indexes of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video respectively;
[0019] Step S24: Determine the relationship between the second abnormality index and the second abnormality index threshold, and when the second abnormality index is greater than or equal to the second abnormality index threshold, add a video abnormality pre-marker to the played video. The playback status of the historical video is associated with the playback abnormality of the first suspected abnormal video, thereby achieving the technical effect of improving the accuracy of determining whether the video file has an abnormality;
[0020] Furthermore, the step S22 of calculating the first abnormality index of each video according to the playback situation includes:
[0021] The total number of times a video is played Pz, the first number of doubtful plays P1, the second number of doubtful plays P2, the third number of doubtful plays P3, the total number of users who play the video Rz, and the number of users with abnormal user identifiers Ry are counted respectively.
[0022] Then the first abnormality index of the video W1=0.63*(0.45*P1+0.2*P2+0.35P3) / Pz+0.37*Ry / Rz,
[0023] When the video is played at a certain time, the ratio of the playing time of the video to the total time of the video is less than or equal to the first ratio threshold, and the video played at this time is the first doubtful play.
[0024] When the video is played at the maximum speed, the video is played at the second playback speed.
[0025] If the link to the previous page of the video played at a certain time belongs to a website other than the video upload website, the video played at that time is considered a third suspicious play. When calculating the first abnormality index, the abnormality of video playback is judged from multiple angles, thereby achieving the technical effect of preventing one-sided evaluation of abnormal playback volume;
[0026] Furthermore, the inspection method further comprises:
[0027] Obtain the total number of times a user has played videos in history Yz, the number of times the user has played videos at the maximum speed Yb, and the number of times when a certain video is played, the ratio of the duration of the video played to the total duration of the video is less than or equal to the second ratio threshold value Yd,
[0028] Calculate the abnormal parameter Yy=0.32*Yb / Yz+0.68*Yd / Yz of the user. If the abnormal parameter is greater than or equal to the abnormal parameter threshold, add an abnormal user ID to the user.
[0029] Furthermore, the inspection method further comprises:
[0030] Count the number of all videos Gz and the number of videos with abnormal pre-marking Gy of the uploader of the first suspected abnormal video, calculate the pre-marking ratio Gb=Gy / Gz, and when the pre-marking ratio is greater than or equal to the pre-marking ratio threshold, add an abnormal upload mark to the uploader.
[0031] Compared with the prior art, the beneficial effect achieved by the present invention is that when detecting whether the playback volume of a certain video is abnormal, the present invention not only detects the playback volume of the video, but also takes other videos of the video uploader of the video as a reference factor for whether the playback volume of the video is abnormal, thereby improving the accuracy of judging whether the playback volume of the video is abnormal. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:
[0033] Figure 1 It is a module schematic diagram of the network abnormal behavior detection system based on big data of the present invention. DETAILED DESCRIPTION
[0034] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0035] See also Figure 1 The present invention provides a technical solution: a network abnormal behavior detection system based on big data, the detection system includes a first suspected abnormal video judgment module and a pre-identification adding judgment module, the first suspected abnormal video judgment module obtains the time interval between the current time and the time when the video file is uploaded when the user detects that the playback volume of a certain video meets certain conditions, and when the time interval is less than or equal to the time interval threshold, the video file is judged to be a first suspected abnormal video, the pre-identification adding judgment module collects the playback status of the historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judges whether to add an abnormal pre-identification to the first suspected abnormal video based on this.
[0036] The first suspected abnormal video judgment module includes a playback volume reference value calculation module and a playback volume reference value comparison module. The playback volume reference value calculation module calculates a playback volume reference value Hc=Hf / Hz according to the total playback volume Hz of the video file and the playback volume Hf of the video file in a preset time period. The playback volume reference value comparison module compares the relationship between the playback volume reference value and the playback volume reference threshold. When the playback volume reference value is greater than or equal to the playback volume reference threshold, the time interval between the current time and the time when the video file is uploaded is obtained; the pre-identification adding judgment module includes a reference video selection module, a first abnormal index calculation module, a second abnormal index calculation module and a second abnormal index comparison module. The reference video selection module collects the playback status of the historical videos of the uploader of the first suspected abnormal video, and compares the historical videos. The videos are sorted in descending order according to the total amount of playback, and the top three videos are selected in turn as the first reference video, the second reference video and the third reference video of the first suspected abnormal video. The first abnormal index calculation module collects the playback conditions of the first suspected abnormal video, the first reference video, the second reference video and the third reference video, and calculates the first abnormal index of each video according to the playback conditions. The second abnormal index calculation module calculates the second abnormal index of the first suspected abnormal video according to the abnormal indexes of the first suspected abnormal video, the first reference video, the second reference video and the third reference video. The second abnormal index comparison module is used to determine the relationship between the second abnormal index and the second abnormal index threshold. When the second abnormal index is greater than or equal to the second abnormal index threshold, a video abnormality pre-marker is added to the playback video.
[0037] The first abnormal index calculation module includes a video playback statistics module and a first abnormal index output module. The video playback statistics module is used to respectively count the total number of playbacks of a certain video Pz, the first doubtful number of playbacks P1, the second doubtful number of playbacks P2, the third doubtful number of playbacks P3, the total number of users Rz who play the video, and the number of users Ry with abnormal user identifiers added. The first abnormal index output module calculates the first abnormal index of the video according to the statistical results of the video playback statistics module. When the video is played at a certain time, the ratio of the playback time of the video to the total time of the video is less than or equal to the first ratio threshold value, and the video played at this time is the first doubtful playback. When the video is played at a certain time, the video is played at the maximum speed, and the video played at this time is the second doubtful playback. If the link to the upper-level page of the video played at a certain time belongs to a website other than the video upload website, the video played at this time is the third doubtful playback.
[0038] The detection system also includes an abnormal user identification adding module and an abnormal upload identification adding module. The abnormal user identification adding module includes an abnormal parameter calculation module and an abnormal parameter comparison module. The abnormal parameter calculation module is used to obtain the total number of historical video playback times Yz of a certain user, the number of times the user plays videos at the maximum speed Yb, and the number of times Yd when a certain video is played, the ratio of the playing time of the video to the total time of the video is less than or equal to the second ratio threshold, and calculate the abnormal parameters of the user based on this. The abnormal parameter comparison module is used to compare the abnormal parameters with the abnormal parameter threshold. When the abnormal parameters are greater than or equal to the abnormal parameter threshold, an abnormal user identification is added to the user. The abnormal upload identification adding module includes a pre-identification ratio calculation module and a pre-identification ratio comparison module. The pre-identification ratio calculation module counts the number of all videos Gz of the uploader of the first suspected abnormal video and the number of videos with abnormal pre-identification Gy, and calculates the pre-identification ratio Gb=Gy / Gz based on this. The pre-identification ratio comparison module compares the pre-identification ratio with the pre-identification ratio threshold. When the pre-identification ratio is greater than or equal to the pre-identification ratio threshold, an abnormal upload identification is added to the uploader.
[0039] A method for detecting abnormal network behavior based on big data, the detection method comprising the following steps:
[0040] Step S1: Obtain the total playback volume Hz of a certain video file and the playback volume Hf of the video file within a preset time period, calculate the playback volume reference value Hc=Hf / Hz, and when the playback volume reference value is greater than or equal to the playback volume reference threshold, obtain the time interval between the current time and the time when the video file is uploaded. If the time interval is less than or equal to the time interval threshold, the video file is the first suspected abnormal video; the behavior of brushing the playback volume often occurs shortly after the video is uploaded and released. Therefore, in this application, the playback volume and the time interval within the preset time period are combined to determine whether it is necessary to further detect the playback volume of the video file;
[0041] Step S2: Collect the playback status of historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judge whether to add an abnormal pre-marker to the first suspected abnormal video based on the information.
[0042] The step S2 further comprises the following:
[0043] Step S21: Collect the playback status of historical videos of the uploader of the first suspected abnormal video, sort the historical videos in descending order according to the total amount of playback, and select the top three videos in turn as the first reference video, the second reference video and the third reference video of the first suspected abnormal video;
[0044] Step S22: collecting playback conditions of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video, and calculating the first abnormality index of each video according to the playback conditions;
[0045] The step S22 of calculating the first abnormality index of each video according to the playback situation includes:
[0046] The total number of times a video is played Pz, the first number of doubtful plays P1, the second number of doubtful plays P2, the third number of doubtful plays P3, the total number of users who play the video Rz, and the number of users with abnormal user identifiers Ry are counted respectively.
[0047] Then the first abnormality index of the video W1=0.63*(0.45*P1+0.2*P2+0.35P3) / Pz+0.37*Ry / Rz,
[0048] When the video is played for a certain time, if the ratio of the playing time of the video to the total playing time of the video is less than or equal to the first ratio threshold, the played video is the first suspicious playing. Generally, when a video starts playing, it is counted as one playing amount. Therefore, when the actual playing time of the video is short, it is possible that it is a malicious playing.
[0049] When the video is played at the maximum speed, the video is played at the second suspicious speed. In actual situations, the fastest speed of video playback is double speed. When the video is played at double speed, the picture speed and voice speed in the video are too fast. Therefore, playing at the maximum speed is not a normal playback state. Therefore, when the video is played at the maximum speed, it may be malicious.
[0050] The link to the parent page of the video played at a certain time, if the link to the parent page belongs to a website other than the website where the video was uploaded, the video played at that time is a third-party playback. The previous page link refers to the link before jumping to the page that plays the video, that is, a user receives a link sent by someone else, and the user jumps to the page that plays the video after clicking the link. Then, the application that user A receives the link sent by others is the link of the previous page that plays the video. If the application is not a video upload website, then the playback is the third suspicious playback. For example, when user A receives a Youku video link sent by someone else on WeChat, user A jumps to the page that plays the Youku video after clicking the link. Then, the playback of the Youku video by user A is the third suspicious playback. In most cases, when users access videos, they browse on the video upload website, and then click on a video to enter the page for watching the video. At this time, the previous page that plays the video belongs to the video website. Therefore, when the previous page of a video does not belong to the video upload website, it may be a malicious playback. The present application also analyzes the user who plays the first suspected abnormal video. When a user has a large number of suspected malicious video playback behaviors in the history of playing videos, it is also very likely that the first suspected abnormal video is played maliciously.
[0051] Step S23: the second abnormality index of the first suspected abnormal video
[0052] Wv=0.72*W1a+0.28*(0.5*W1b+0.3*W1c+0.2*W1d),
[0053] Wherein, W1a, W1b, W1c, and W1d are the abnormality indexes of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video respectively;
[0054] Step S24: Determine the relationship between the second abnormality index and the second abnormality index threshold, and when the second abnormality index is greater than or equal to the second abnormality index threshold, add a video abnormality pre-marker to the played video.
[0055] The inspection method further comprises:
[0056] Count the number of all videos Gz and the number of videos with abnormal pre-marks Gy of the uploader of the first suspected abnormal video, calculate the pre-mark ratio Gb=Gy / Gz, and when the pre-mark ratio is greater than or equal to the pre-mark ratio threshold, add an abnormal upload mark to the uploader. Adding an abnormal upload mark to the video uploader makes it easier for video playback supervisors to control video uploaders with more video playback abnormalities;
[0057] The inspection method further comprises:
[0058] Obtain the total number of times a user has played videos in history Yz, the number of times the user has played videos at the maximum speed Yb, and the number of times when a certain video is played, the ratio of the duration of the video played to the total duration of the video is less than or equal to the second ratio threshold value Yd,
[0059] Calculate the abnormal parameter Yy=0.32*Yb / Yz+0.68*Yd / Yz of the user. If the abnormal parameter is greater than or equal to the abnormal parameter threshold, add an abnormal user ID to the user.
[0060] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.
[0061] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. 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 network abnormal behavior detection system based on big data, characterized in that: The detection system includes a first suspected abnormal video judgment module and a pre-mark adding judgment module. When the user of the first suspected abnormal video judgment module detects that the playback volume of a certain video meets certain conditions, the module obtains the time interval between the current time and the time when the video file is uploaded. When the time interval is less than or equal to the time interval threshold, the video file is judged to be a first suspected abnormal video. The pre-mark adding judgment module collects the playback status of the historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judges whether to add an abnormal pre-mark to the first suspected abnormal video based on the collected information; The detection method includes the following steps: Step S1: When the playback volume of a certain video meets certain conditions, the time interval between the current time and the time when the video file is uploaded is obtained. If the time interval is less than or equal to the time interval threshold, the video file is the first suspected abnormal video. Step S2: collecting the playback status of historical videos of the uploader of the first suspected abnormal video and the playback status of the first suspected abnormal video each time within a preset time period, and judging whether to add an abnormal pre-marker to the first suspected abnormal video based on the above; When the playback volume of a certain video meets certain conditions, step S1 includes the following: Get the total playback volume Hz of a video file and the playback volume Hf of the video file in a preset time period, calculate the playback volume reference value Hc=Hf / Hz, and when the playback volume reference value is greater than or equal to the playback volume reference threshold, get the time interval between the current time and the time when the video file was uploaded; The step S2 further comprises the following: Step S21: Collect the playback status of historical videos of the uploader of the first suspected abnormal video, sort the historical videos in descending order according to the total amount of playback, and select the top three videos in turn as the first reference video, the second reference video and the third reference video of the first suspected abnormal video; Step S22: collecting playback conditions of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video, and calculating the first abnormality index of each video according to the playback conditions; Step S23: the second abnormality index of the first suspected abnormal video Wv=0.72*W1a+0.28*(0.5*W1b+0.3*W1c+0.2*W1d), Wherein, W1a, W1b, W1c, and W1d are first abnormality indexes of the first suspected abnormal video, the first reference video, the second reference video, and the third reference video, respectively; Step S24: determining the relationship between the second abnormality index and the second abnormality index threshold, and when the second abnormality index is greater than or equal to the second abnormality index threshold, adding a video abnormality pre-marker to the first suspected abnormal video; The step S22 of calculating the first abnormality index of each video according to the playback situation includes: The total number of times a video is played Pz, the first number of doubtful plays P1, the second number of doubtful plays P2, the third number of doubtful plays P3, the total number of users who play the video Rz, and the number of users with abnormal user identifiers Ry are counted respectively. Then the first abnormality index of the video W1=0.63*(0.45*P1+0.2*P2+0.35*P3) / Pz+0.37*Ry / Rz, When the video is played at a certain time, the ratio of the playing time of the video to the total time of the video is less than or equal to the first ratio threshold, and the video played at this time is the first doubtful play. When the video is played at the maximum speed, the video is played at the second playback speed. The link to the previous page of the video played at a certain time, if the previous page link belongs to a website other than the website where the video was uploaded, the video played at that time is a third-party play; The detection method further comprises: Obtain the total number of times a user has played videos in history Yz, the number of times the user has played videos at the maximum speed Yb, and the number of times when a certain video is played, the ratio of the duration of the video played to the total duration of the video is less than or equal to the second ratio threshold value Yd, Calculate the abnormal parameter Yy=0.32*Yb / Yz+0.68*Yd / Yz of the user. If the abnormal parameter is greater than or equal to the abnormal parameter threshold, add an abnormal user ID to the user. The detection method further comprises: Count the number of all videos Gz and the number of videos with abnormal pre-marking Gy of the uploader of the first suspected abnormal video, calculate the pre-marking ratio Gb=Gy / Gz, and when the pre-marking ratio is greater than or equal to the pre-marking ratio threshold, add an abnormal upload mark to the uploader.
