CN110225373B - Video auditing method and device and electronic equipment - Google Patents

Video auditing method and device and electronic equipment Download PDF

Info

Publication number
CN110225373B
CN110225373B CN201910509756.1A CN201910509756A CN110225373B CN 110225373 B CN110225373 B CN 110225373B CN 201910509756 A CN201910509756 A CN 201910509756A CN 110225373 B CN110225373 B CN 110225373B
Authority
CN
China
Prior art keywords
video
scheduling
audited
factor
sorting
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910509756.1A
Other languages
Chinese (zh)
Other versions
CN110225373A (en
Inventor
刘刚
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tencent Technology Shenzhen Co Ltd
Original Assignee
Tencent Technology Shenzhen Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tencent Technology Shenzhen Co Ltd filed Critical Tencent Technology Shenzhen Co Ltd
Priority to CN201910509756.1A priority Critical patent/CN110225373B/en
Publication of CN110225373A publication Critical patent/CN110225373A/en
Application granted granted Critical
Publication of CN110225373B publication Critical patent/CN110225373B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/23—Processing of content or additional data; Elementary server operations; Server middleware
    • H04N21/239—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests
    • H04N21/2393—Interfacing the upstream path of the transmission network, e.g. prioritizing client content requests involving handling client requests
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
    • H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs
    • H04N21/44008—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/45—Management operations performed by the client for facilitating the reception of or the interaction with the content or administrating data related to the end-user or to the client device itself, e.g. learning user preferences for recommending movies, resolving scheduling conflicts
    • H04N21/454—Content or additional data filtering, e.g. blocking advertisements
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
    • H04N21/47—End-user applications
    • H04N21/475—End-user interface for inputting end-user data, e.g. personal identification number [PIN], preference data
    • H04N21/4751—End-user interface for inputting end-user data, e.g. personal identification number [PIN], preference data for defining user accounts, e.g. accounts for children
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
    • H04N21/80—Generation or processing of content or additional data by content creator independently of the distribution process; Content per se
    • H04N21/83—Generation or processing of protective or descriptive data associated with content; Content structuring
    • H04N21/835—Generation of protective data, e.g. certificates
    • H04N21/8352—Generation of protective data, e.g. certificates involving content or source identification data, e.g. Unique Material Identifier [UMID]

Abstract

The invention relates to the technical field of computers, in particular to a video auditing method, a device and electronic equipment, wherein the method comprises the steps of filtering out videos to be audited, the contents of which accord with forbidden content conditions, according to the contents of each video to be audited; aiming at each filtered video to be audited, scheduling priority ranking is carried out according to at least one scheduling ranking factor of the video to be audited; according to the scheduling priority ranking, re-auditing resources are preferentially distributed to the videos to be audited with high ranking, so that the accuracy and the efficiency can be improved, overstock of some time-effect or high-quality videos can be avoided, the videos needing to be audited preferentially can be preferentially distributed with the re-auditing resources, and the overall performance of video auditing is improved.

Description

Video auditing method and device and electronic equipment
Technical Field
The invention relates to the technical field of computers, in particular to a video auditing method and device and electronic equipment.
Background
In the era of rapid development of the internet, a plurality of websites support and allow users to upload videos and display the videos to the public, the video uploading amount rapidly increases at an exponential speed along with the reduction of the threshold of content production, and in order to ensure the security of distributed content, the verification of the video content needs to be completed in a short time, for example, whether the content relates to sensitive information, content quality, security and the like is identified and processed, at present, manual verification is mainly relied on, the efficiency is low, and a large amount of labor cost needs to be consumed.
The prior art also provides a video auditing method, which mainly only audits video contents, firstly filters videos obviously violating legal contents through a machine algorithm, prompts auditing results of the machine algorithm to auditors, then leads the auditors to get orders according to a time sequence, and audits other videos which are not filtered again.
Disclosure of Invention
The embodiment of the invention provides a video auditing method and device and electronic equipment, and aims to solve the problem that in the prior art, the efficiency and accuracy of video content auditing are low.
The embodiment of the invention provides the following specific technical scheme:
one embodiment of the invention provides a video auditing method, which comprises the following steps:
filtering the video to be audited with the picture content meeting the forbidden content condition according to the picture content of each video to be audited;
aiming at each filtered video to be audited, carrying out scheduling priority ranking according to at least one scheduling ranking factor of the video to be audited;
and according to the scheduling priority level sequence, allocating re-audit resources for the videos to be audited with high level sequence preferentially.
Another embodiment of the present invention provides a video auditing apparatus, including:
the filtering module is used for filtering the videos to be audited, the picture contents of which accord with the forbidden content conditions, according to the picture contents of the videos to be audited;
the sorting module is used for sorting the scheduling priority levels of each filtered video to be audited according to at least one scheduling sorting factor of the video to be audited;
and the distribution module is used for distributing re-audit resources to the videos to be audited with high rank according to the scheduling priority rank ordering.
In combination with another embodiment of the present invention, when performing scheduling priority ranking according to at least one scheduling ranking factor of a video to be audited, the ranking module is specifically configured to:
determining the value of each scheduling sorting factor according to the characteristic information of each scheduling sorting factor in each scheduling sorting factor aiming at each filtered video to be audited;
determining a scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor;
and according to the determined scheduling sorting score, performing scheduling priority sorting on each filtered video to be audited.
With reference to another embodiment of the present invention, if the scheduling ordering factors include at least two, when determining the scheduling ordering score of the video to be audited according to the value of each scheduling ordering factor, the ordering module is specifically configured to: and determining the scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor and the influence weight corresponding to each scheduling sorting factor.
In combination with another embodiment of the present invention, the at least one scheduling ordering factor is set according to an element associated with timeliness of video review; the at least one scheduling ordering factor comprises one or any combination of the following: video account quality, video hotspot level, video cover definition level.
With reference to another embodiment of the present invention, if the at least one scheduling ranking factor is the quality of the video account, and when the value of each scheduling ranking factor is determined according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, the ranking module is specifically configured to:
determining a video account corresponding to a video to be audited, and acquiring the total number of uploaded videos, the audit result of each uploaded video and the playing times of each audited video of the corresponding video account in a preset time period;
determining the auditing passing rate of the video account according to the total number of the uploaded videos and the auditing result of each uploaded video;
determining account influence of the video account according to the playing times of the videos passing the verification;
and determining the quality of the video account according to the auditing passing rate and the account influence.
In combination with another embodiment of the present invention, if the at least one scheduling ordering factor is a video hotspot level, and when the value of each scheduling ordering factor is determined according to the characteristic information of each scheduling ordering factor in each scheduling ordering factor, the ordering module is specifically configured to:
acquiring a label of a video to be audited;
comparing the tags of the video to be audited with the pre-configured hot spot tags according to the pre-configured hot spot tags;
and determining the video hotspot grade of the video to be audited according to the number of successfully compared and matched labels of the video to be audited and the hotspot labels.
In combination with another embodiment of the present invention, if the at least one scheduling ordering factor is a video cover definition level, and when a value of each scheduling ordering factor is determined according to feature information of each scheduling ordering factor in each scheduling ordering factor, the ordering module is specifically configured to:
acquiring a video cover of a video to be audited;
and respectively obtaining the definition grades determined by taking the video covers as input parameters based on a definition recognition model, wherein the definition recognition model is used for determining the definition grades according to the video covers.
In combination with another embodiment of the present invention, further comprising:
the acquisition module is used for acquiring the auditing result of the re-auditing resource;
and the updating module is used for updating the prohibited content conditions and/or updating the influence weight corresponding to each scheduling sequencing factor according to the auditing result of the re-auditing resource.
Another embodiment of the present invention provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program according to any of the steps of the video auditing method.
Another embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, performs the steps of any one of the video auditing methods described above.
In the embodiment of the application, when video auditing is performed, videos to be audited, of which the picture contents meet the conditions of forbidden contents, are filtered, then multi-aspect dimension information is considered, scheduling priority ranking is performed according to at least one scheduling ranking factor of the videos to be audited aiming at each filtered video to be audited, re-auditing resources are preferentially allocated to the videos to be audited, of which the ranking is high, so that only the video contents are considered during the video auditing, the accuracy is improved, after the videos are preliminarily audited and filtered, the scheduling priority ranking is performed on each filtered video to be audited, the re-auditing resources are allocated according to the scheduling priority ranking, the backlog of some time-efficiency or high-quality videos can be avoided, the video needing the preferential auditing resources can be preferentially allocated to the re-auditing resources, the auditing efficiency of time efficiency or high-quality videos is greatly improved, the time consumption of auditing is reduced, and the overall performance of the video auditing is improved.
Drawings
FIG. 1 is a schematic diagram of an application architecture of a video auditing method in an embodiment of the application;
FIG. 2 is a flow chart of a video review method in an embodiment of the present application;
FIG. 3 is a block diagram of a video review system in an embodiment of the present application;
fig. 4 is a schematic flowchart of a video uploading process in an embodiment of the present application;
FIG. 5 is a schematic flow chart of a video preliminary review and scheduling sorting process in an embodiment of the present application;
fig. 6 is a schematic flow chart of a video scheduling and distributing process in the embodiment of the present application;
FIG. 7 is a schematic view of a process flow of manual video review in an embodiment of the present application;
fig. 8 is a schematic flow chart of a video distribution process in the embodiment of the present application;
fig. 9 is a schematic structural diagram of a video auditing apparatus according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
In order to facilitate an understanding of the embodiments of the present invention, a few concepts are briefly introduced below:
video: in the embodiment of the present Application, mainly, the Application programs, such as a QQ viewpoint, a short video Application (APP), and the like, are recommended to the user for reading, and may include a vertical version of a small video, a horizontal version of a short video, and the like, and are mainly provided in the form of information streams (Feeds).
Short video: the short video is an internet content transmission mode, generally is a video transmission content which is transmitted on a new internet medium for a period of less than 5 minutes, and with the popularization of mobile terminals and the acceleration of networks, the short video gradually gains favor of various large platforms, fans and capital.
Feeds: the message sources, through which the website can propagate the latest information to users, are usually arranged in a Timeline (Timeline), where Timeline is the most primitive and basic presentation form of Feed, and a prerequisite for a user to subscribe to the website is that the website provides the message sources, feed is converged together and called aggregation, and software for aggregation is called aggregator, which is software dedicated to subscribe to the website for an end user and may be generally called Simple Syndication (RSS) reader, feed reader, news reader, and the like.
Professional Generated Content (PGC): is an internet term that refers to professionally produced content, such as video websites, expert produced content, such as micro blogs, etc.
Multi-Channel Network (MCN): the method is a product form of a multi-channel network, combines PGC (product content control) contents, and ensures continuous output of the contents under the powerful support of capital, thereby finally realizing stable business achievement.
User Generated Content (UGC): the creation of the Web2.0 concept along with the promotion of personalization is not a specific service, but a new way for users to use the Internet, namely downloading and uploading from the original downloading to the main change.
Professional User authored Content (PUGC): in the form of UGC, the produced professional audio-video content is relatively close to that of PGC.
Machine learning: the method is a multi-field interdiscipline, relates to a plurality of disciplines such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like, and is used for specially researching how a computer simulates or realizes the learning behavior of human beings so as to obtain new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer.
Deep learning: the concept of deep learning is derived from the research of an artificial neural network, and a multi-layer perceptron with multiple hidden layers is a deep learning structure, and the deep learning forms more abstract high-layer representation attribute categories or features by combining low-layer features so as to find distributed feature representations of data.
With the rapid development of the internet, many websites support and allow users to upload videos and display the videos to the public by themselves, with the decrease of the threshold of content production, the amount of video uploading is rapidly increasing at an exponential rate, such as PGC, UGC content and the like from media and organizations, and with the development of networks, the demand of users for receiving information gradually trends to the video era, especially short videos, which are usually displayed in a form of Feeds stream for users to refresh quickly, so that short videos are uploaded from the UGC, PGC, users to organizations specially manufacturing short videos, to MCN, to professional short video apps and other flow platforms, and short videos have become one of important transmission modes.
The prior art also provides a mode, a large amount of manpower is used for simultaneously assisting in video auditing by machine algorithm capability, mainly, the auditing is only carried out on video content per se, videos obviously violating laws are filtered out by the machine algorithm, auditing results of the machine algorithm are prompted to auditors, the auditors get orders according to time sequence, and the other videos which are not filtered out are audited again.
Therefore, in view of the above problems, an embodiment of the present application provides a video auditing method, where a video to be audited is subjected to preliminary auditing according to picture content of the video to be audited, a video to be audited whose picture content meets a prohibited content condition is filtered, then, for each filtered video to be audited, scheduling priority ranking is performed according to at least one scheduling ranking factor of the video to be audited, and re-auditing resources are preferentially allocated to the video to be audited whose ranking is high according to the scheduling priority ranking, so that a part of the video to be audited whose picture content meets the prohibited content condition is filtered first, then, multi-aspect dimension information is considered, a scheduling ranking factor is calculated, each filtered video to be audited is subjected to scheduling priority ranking, each filtered video to be audited can be distributed according to the scheduling priority ranking, accuracy and efficiency are higher, for the video with higher scheduling priority ranking, auditing efficiency is greatly improved, backlog of the video with high quality or high timeliness is avoided, manpower viewing experience of the video displayed by a user is improved, a large amount of auditing cost is saved, and learning efficiency of the videos with higher auditing efficiency and higher auditing efficiency of the respective machine can be considered, and the efficiency of the high-time delay of the video can be reduced, and the efficiency of the user can be greatly reduced, and the time delay of the user can be enabled.
Fig. 1 is a schematic diagram of an application architecture of a video auditing method according to an embodiment of the present invention, including a terminal 100 and a server 200.
The terminal 100 may be any intelligent device such as a smart phone, a tablet computer, and a portable personal computer, and various APPs may be installed on the terminal 100, for example, the APP installed on the terminal 100 in this embodiment may be applications of various acceptable messages and Feeds stream information such as QQ and wechat, and a user may upload a video through the APP, and the APP may also receive and display a video that is recommended by the server 200 and passed through verification, so that the user may browse and watch the video.
The server 200 can provide various network services for the terminal 100, and for different applications on the terminal 100, the server 200 may be considered as a background server providing corresponding network services, for example, in the embodiment of the present invention, the server 200 may receive a video uploaded by the terminal 100, perform video review, and return a result of the video review, and for example, the server 200 may recommend various videos that have been reviewed to the terminal 100 for a user to watch.
The server 200 may be a server, a server cluster formed by a plurality of servers, or a cloud computing center.
Specifically, the server 200 may include a processor 210 (CPU), a memory 220, an input device 230, an output device 240, and the like, the input device 230 may include a keyboard, a mouse, a touch screen, and the like, and the output device 240 may include a Display device, such as a Liquid Crystal Display (LCD), a Cathode Ray Tube (CRT), and the like.
Memory 220 may include Read Only Memory (ROM) and Random Access Memory (RAM), and provides processor 210 with program instructions and data stored in memory 220. In an embodiment of the present invention, the memory 220 may be used for storing a program of any one of the video auditing methods in the embodiment of the present invention.
The processor 210 is configured to execute the steps of any of the video auditing methods according to the embodiments of the present invention by calling the program instructions stored in the memory 220 and the processor 210 is configured to execute the steps according to the obtained program instructions.
The terminals 100 and the server 200 are connected via the internet to communicate with each other. Optionally, the internet described above uses standard communication techniques and/or protocols. The internet is typically the internet, but can be any Network including, but not limited to, a Local Area Network (LAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), any combination of mobile, wireline or wireless networks, private or virtual private networks. In some embodiments, data exchanged over a network is represented using techniques and/or formats including Hypertext Mark-up Language (HTML), extensible Markup Language (XML), and the like. All or some of the links may also be encrypted using conventional encryption techniques such as Secure Socket Layer (SSL), transport Layer Security (TLS), virtual Private Network (VPN), internet Protocol Security (IPsec). In other embodiments, custom and/or dedicated data communication techniques may also be used in place of, or in addition to, the data communication techniques described above.
It should be noted that the application architecture diagram in the embodiment of the present invention is for more clearly illustrating the technical solution in the embodiment of the present invention, and does not limit the technical solution provided in the embodiment of the present invention, and is not limited to short videos, and the technical solution provided in the embodiment of the present invention is also applicable to similar problems for other application architectures and business applications.
It should be noted that, in each embodiment of the present invention, the application architecture shown in fig. 1 is taken as an example to schematically illustrate the application of the video auditing method.
Based on the foregoing embodiment, a video auditing method in an embodiment of the present invention is described below, and referring to fig. 2, a flowchart of a video auditing method in an embodiment of the present invention is shown, where the method includes:
step 200: and filtering the video to be audited, the picture content of which accords with the forbidden content condition, according to the picture content of each video to be audited.
In practice, for each video uploaded by a user, since the video content uploaded by the user is various, and therefore needs to be audited, in this embodiment of the present application, a preliminary audit is performed according to the picture content, and a to-be-audited video whose picture content meets the prohibited content condition is filtered out, specifically, the step 200 is executed, including:
1) And respectively extracting key frames of the videos to be audited.
2) And respectively extracting the picture content of each key frame, comparing the picture content with the prohibited content conditions, determining the video to be audited which accords with the prohibited content conditions as the prohibited video, and filtering the video to be audited which is determined as the prohibited video.
Since the video is a picture composed of pictures and audio and taking a frame as a unit, the pictures, audio and characters in the picture content can be audited during preliminary audit.
For example, for the audio which may have some bad sounds such as violence, obscene, and the like, the audio in the picture content of each key frame may be extracted, identified, compared with the prohibited audio condition, and if the prohibited audio condition is met, the video to be audited is filtered.
For another example, for the recognition and verification of pictures or characters in the picture content, the extracted image features are mainly compared with prohibited content conditions based on an image recognition technology, wherein the prohibited content conditions may be set according to actual situations, for example, violent content, pornographic content and the like may be included generally, so that when verification is performed, whether some star characters exist or not may be judged through a face recognition technology, and whether pornographic pictures exist or not may also be judged through different dimensions such as normal, sexual feeling, pornographic and the like by recognizing the naked state of the picture. In addition, for the recognition of the characters in the picture, a character recognition (OCR) technology or the like may be used, and the embodiment of the present application is not limited.
Further, during the preliminary review, the definition of the picture content of the key frame may also be determined, and videos to be reviewed with too low definition are filtered out, for example, the definition level is divided into 10 grades, and videos to be reviewed with the definition level less than or equal to 2 grades may be filtered out.
Therefore, through the preliminary audit, the videos to be audited, of which the picture contents meet the forbidden content conditions, are filtered, some videos to be audited, of which the picture contents are harmful or bad, can be filtered out firstly, the workload of subsequent recheck is reduced, the efficiency is improved, and the accuracy of the audit can be improved by performing subsequent recheck only on the videos to be audited, of which the picture contents are not filtered.
Step 210: and aiming at each filtered video to be audited, carrying out scheduling priority ranking according to at least one scheduling ranking factor of the video to be audited.
The scheduling ranking factor is set according to an element associated with timeliness of video auditing, that is, the scheduling ranking factor is set according to a priority order of the video auditing, where timeliness refers to the priority order of the video auditing, and the setting of the scheduling ranking factor is determined based on timeliness consideration dimensionality, for example, if timeliness consideration dimensionality is mainly definition, and definition is higher and needs to be prioritized, the scheduling ranking factor can be set to a definition level.
Step 220: and according to the scheduling priority level sequence, allocating re-audit resources for the videos to be audited with high level sequence preferentially.
The re-auditing resource allocated in the embodiment of the application can be a manual service resource, for example, videos to be audited with high rank ordering are preferentially distributed to manual services according to scheduling priority rank ordering, so that the manual services perform re-auditing on the videos to be audited, that is, whether the videos to be audited belong to prohibited videos or not are judged, whether the videos need to be filtered or not is judged, the filtered videos cannot be distributed to audiences, and videos which pass the auditing can be distributed to the audiences if the videos are not filtered.
Furthermore, after the video to be audited is allocated to the resources to be audited again, a possible implementation mode is provided, and the auditing result of the resources to be audited again is obtained; and updating the forbidden content condition and/or updating the influence weight corresponding to each scheduling sorting factor according to the auditing result of the re-auditing resource.
That is to say, in the embodiment of the present application, the review result of the re-review resource, for example, the review result of the manual review, may be read, and the video review method may be updated according to the review result of the manual review, for example, the prohibited content condition and the influence weight of the scheduling order factor are updated, and the system video review method and the manual review are combined, so that the video review accuracy may be improved, the performance of the system video review method may be improved, and thus the efficiency may be effectively improved, and the subsequent manual review cost may be saved.
Of course, the review resource may also be other resources, and the embodiment of the present application is not limited.
A specific implementation method of step 210 in the above embodiment is described below.
When step 210 is executed, for each filtered video to be audited, scheduling priority ranking is performed according to at least one scheduling ranking factor of the video to be audited, which specifically includes:
s1, aiming at each filtered video to be audited, determining the value of each scheduling sorting factor according to the characteristic information of each scheduling sorting factor in each scheduling sorting factor.
Wherein, at least one scheduling ordering factor comprises one or any combination of the following: the quality of the video account, the video hotspot level and the video cover definition level are not limited in the embodiment of the application, and other scheduling and sequencing factors can be set according to the auditing requirement, so that the expected video to be audited can be audited preferentially.
According to different scheduling ordering factors, there may be the following cases when S1 is specifically executed:
in the first case: if at least one scheduling ranking factor is the quality of the video account, determining the value of each scheduling ranking factor according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, specifically comprising:
1) Determining a video account corresponding to a video to be audited, and acquiring the total number of uploaded videos, the auditing result of each uploaded video and the playing times of each audited video in a preset time period of the corresponding video account.
The preset time period, for example, a certain time period before the video auditing method is executed, for example, within the previous month, and the like, is not limited.
In the embodiment of the application, it is considered that all videos are finally provided by content producers, good videos should correspond to the good content producers, videos uploaded by the good content producers should be preferentially checked during checking and scheduling, and the priority of checking and scheduling should be reduced for low-level content producers relatively, so that the quality of a video account can be used as a scheduling ranking factor in the embodiment of the application, and in evaluating the quality of the video account, various dimensional characteristics such as checking pass rate, account influence and the like can be adopted.
2) And determining the auditing passing rate of the video account according to the total number of the uploaded videos and the auditing result of each uploaded video.
For example, for the audit passing rate of the video account, the audit flow data of the last month can be obtained, and the following field information is used: rowkey: unique identification of the published video content, pu: and issuing the unique identification, the auditing date and the auditing result of the video account of the video content, wherein the auditing result comprises whether the video account passes or not, if not, the reason of failing, so that the total uploaded video number of the video account corresponding to each video to be audited and the auditing result of each uploaded video are obtained.
Specifically, the audit passing rate of the video account is calculated by the following method:
Figure GDA0003970688790000121
wherein, s1 i Indicates the audit pass rate, o i,j A flag indicating whether the ith account number and the jth content are approved or not, 1 indicates pass, 0 indicates fail, and t indicates pass 0 Indicates the current date, t i,j Indicating the ith account number, the date on which the jth content was reviewed, δ, η indicating the control parameter for smoothing, e.g., the suggested values δ =1.0, η =10.0.
3) And determining account influence of the video account according to the playing times of the videos passing the audit.
For example, for account influence of a video account, the following field information may be utilized according to running data of video playing times of the last month: and (3) Pu in: the unique identification and date of the video account for issuing the video content, and the playing times generated by distribution after the video account is enabled are checked.
Specifically, the account influence of the video account is calculated in the following manner:
s2 i ＝∑ j σ(v i,j )/(t 0 -t i,j +δ)
wherein, s2 i Representing account influence, v i,j Indicates the ith account number, the playing times of the jth day, t 0 Indicating the current date, t i,j Indicating the ith account number, the day of the jth day, delta indicating the control parameter of the smoothing action, e.g. the suggested value is delta =1.0, sigma (·) indicating the function of the smoothing action, e.g. the suggested expression is
Figure GDA0003970688790000131
4) And determining the quality of the video account according to the auditing passing rate and the account influence.
For example, the video account quality may be calculated by: s i ＝(1.0+s1 i )(1.0+s2 i ) 0.5 。
In the embodiment of the application, it can be considered that the lower the review passing rate is, the poorer the quality of the corresponding video account is, but it is considered that the uploading video amount of each video account is not uniform, so that the influence of the video account is also considered, the larger the influence of the account is, the higher the quality of the video account is, the quality of the video account is comprehensively evaluated through the review passing rate and the influence of the account, the higher the quality of the video account is, the quality of the video account is represented as a high-quality account, and otherwise, the lower the quality of the video account is, the quality of the video account is represented as a low-quality account.
Certainly, in the embodiment of the application, the quality of the video account is not limited to the factor of the passing rate of the audit and the influence of the account, and other factors may also be considered to evaluate the quality of the video account, for example, the fan activity of the video account and the like may also be utilized, some positive factors may improve the quality of the video account, and some negative factors, such as negative feedback and report information and the like of some users, may reduce the quality of the video account, and may be specifically set according to actual needs, and is not limited.
In the second case: if at least one scheduling ranking factor is a video hotspot level, determining the value of each scheduling ranking factor according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, which specifically comprises the following steps:
1) And acquiring a label of the video to be audited.
In the embodiment of the application, some video contents related to the hotspot event generally have high auditing priority so as to ensure that the hotspot video contents can be audited preferentially and reduce auditing backlog and delay of the hotspot video, so that the video hotspot level can be used as a scheduling sorting factor.
Each video usually has a corresponding title and a Tag (Tag), and the Tag may be selected by a video uploader or extracted by the system according to the content of the video, so as to obtain tags of videos to be audited, where at least one Tag of a video to be audited is provided.
2) And comparing the tag of the video to be audited with each pre-configured hotspot tag according to each pre-configured hotspot tag.
The pre-configured hot spot labels can be used for configuring related events and labels by operators according to experience, and the hot spot labels can be manually referred to by crawling of various portal websites, such as microblogs, search engines and the like.
3) And determining the video hotspot grade of the video to be audited according to the number of successfully compared and matched labels of the video to be audited and each hotspot label.
Specifically, the number of successful comparison and matching is determined, and the video hotspot level of the video to be audited is determined according to the mapping relation between the number of successful comparison and matching and the video hotspot level.
For example, the video hotspot level is divided into 10 levels, each level is correspondingly provided with the number of hit hotspot tags, and the more hotspot tags hit by a certain video to be audited, the higher the corresponding video hotspot level is.
In the third case: if at least one scheduling ranking factor is a video cover definition level, determining the value of each scheduling ranking factor according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, wherein the method specifically comprises the following steps:
1) And acquiring a video cover of the video to be audited.
In the embodiment of the application, the video cover is considered to be the first feeling of a user seeing a video, if the definition quality is poor, the direct impression is that the quality of video content is poor, the desire of continuous clicking is avoided, and the watching experience of the user is reduced, so that the video cover is used as a scheduling ranking factor, and the higher the definition grade of the video cover is, the prior auditing is required.
The video cover map can be selected by the user when the user uploads the video, or can be extracted by the system, for example, if the user does not select the video cover when uploading the video, the system can randomly extract one frame of picture from the video to be used as the video cover.
2) And respectively obtaining the definition levels determined by taking the video covers as input parameters based on the definition recognition model, wherein the definition recognition model is used for determining the definition levels according to the video covers.
The definition recognition model is obtained by pre-training, a supervised learning labeling sample and a deep learning method can be adopted, for example, the definition grade can be divided into 10 grades, a video cover mainly comes from a frame extraction of video content, a key frame can be extracted, priori knowledge is fused, definition labeling is carried out on the key frame, a training sample is obtained, when the definition recognition model is trained, the training sample is input into the definition recognition model, the definition grade is output, a loss function of training is modified by combining a classification problem and a regression problem, an output result accords with the subjective meaning of the definition grade, in addition, when the definition recognition model is trained, other characteristic information of the video cover can be considered, for example, the width and the height of a picture, the number of stored bytes, a cartoon picture, background blurring, special adaptation to a vertical screen and the like.
Therefore, based on the trained definition recognition model, the definition level of the video cover of the video to be audited is obtained, and the higher the definition level is, the higher the audit priority should be.
And S2, determining the scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor.
Further, if the scheduling ranking factors include at least two, determining a scheduling ranking score of the video to be audited according to the value of each scheduling ranking factor, specifically including: and determining the scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor and the influence weight corresponding to each scheduling sorting factor.
For example, the scheduling ranking factor includes video account quality, video hotspot level and video cover definition level, and the scheduling ranking score of the video to be reviewed may be calculated as follows:
Figure GDA0003970688790000151
where F is the scheduling order score, in the first parenthesis
Figure GDA0003970688790000152
Represents the audit pass rate, ∑ P pass Checking the number of passed videos, sigma P total For the total upload number, 3 in the numerator and 5 in the denominator are smoothing parameters, such as bayesian smoothing, the multiplication of the first parenthesis and the second parenthesis represents the quality of the video account, and the index of each parenthesis in the formula is an influence weight, which can be determined empirically or in other ways, and is not limited.
Because the dimensions of different scheduling ranking factors are different, a multiplication method is adopted among the scheduling ranking factors when the scheduling ranking score is calculated in the embodiment of the application.
Further, the first parenthesis and the second parenthesis in the above formula F may be replaced by the way of calculating the quality of the video account in the above embodiment, that is, F may be replaced by F
Figure GDA0003970688790000161
Figure GDA0003970688790000162
Is replaced by s i ，s i ＝(1.0+s1 i )(1.0+s2 i ) 0.5 。
Further, when each scheduling ordering factor is set in the embodiment of the present application, a test needs to be performed in advance to verify whether a single scheduling ordering factor is effective or not and how large the effect is affected, and in order to measure and adjust the effect of each scheduling ordering factor and adjust the effect to the optimum, specifically, in the embodiment of the present application, a video content experiment bucket based on the audit flow and the video content warehousing flow is introduced, the scheduling ordering scores can be calculated and ordered based on the single scheduling ordering factor respectively, and the videos to be audited and ordered based on different single scheduling ordering factors are placed in different video content experiment buckets respectively for subsequent allocation of re-audit resources.
For example, if each auditor can audit about 800 videos per day, 1000 auditors are needed if 80W videos need to be audited per day, and auditors usually receive audition tasks in batches, for example, 20 videos at a time, in order to ensure efficiency. Because the scheduling ordering factors are many, in order to measure the effect of each scheduling ordering factor individually, a/B test (abest) verification needs to be performed under the condition that other factors are not changed, at this time, a certain number of audits needs to be ensured for an experimental group and a comparison group, the statistical relevance of the audits is too small, the statistical relevance of the audits is not obvious, the audits are too large, and whether the allocated groups are too small is not enough, and a plurality of groups of experiments cannot be performed, therefore, in the embodiment of the application, 80W videos per day are divided into 100 groups, that is, 100 groups of video content experiment buckets are arranged, 8000 auditing tasks in each group correspond to 400 auditing receipts, where a specific auditing task enters auditing and is determined by a content scheduling and bucket service according to configuration rules, for example, the video content experiment bucket No. 1 is scheduled and ordered by adding video account quality, 1000 auditors acquire auditing tasks from 100 buckets in a random manner, obtaining a plurality of audit tasks, for example 20 audit tasks, from one bucket at a time, marking the audit tasks which are obtained from which video content experiment bucket, further obtaining an audit result of manual audit, and adjusting the influence weight of the corresponding scheduling ranking factor according to the audit result of manual audit, for example, the low-quality account content in the filtering rejection reason is high, at this time, the weight of the video account quality can be improved, for example, the hot content ex-warehouse experiment is high, the video hot rank weight can be increased, so that when the comparison bucket is fused with the passing rate of a single scheduling ranking factor, after the scheduling parameter is adjusted, the passing rate is not improved or not reduced, after the scheduling effect is stable, the influence weight of each scheduling ranking factor can be obtained, and then, after the video to be audited is ranked according to the priority of a plurality of scheduling ranking factors, all video content experiment buckets can be uniformly entered from high to low according to the priority level, and the distribution of re-auditing resources is waited.
And S3, according to the determined scheduling sorting score, performing scheduling priority sorting on each filtered video to be audited.
Specifically, the higher the scheduling ranking score is, the higher the scheduling priority level of the corresponding video to be reviewed is, and each filtered video to be reviewed may be sequentially subjected to scheduling priority level ranking from high to low according to the scheduling ranking score.
In the embodiment of the application, aiming at each filtered video to be audited, the scheduling sorting score is calculated based on the scheduling sorting factor considered in video auditing, so that scheduling priority sorting is performed, rechecking resources are distributed, instead of auditing by auditors according to a time sequence, the starting amount of effective qualified video content can be greatly increased, the accuracy and the efficiency are improved, some time-efficient or high-quality videos can be audited preferentially, the processing efficiency is improved, the whole video content ecology is facilitated, and the watching experience of a user on the displayed videos passing the auditing is also improved.
Based on the above embodiments, the following further describes the video auditing method in the embodiments of the present invention with specific application scenarios. Taking review resources as manual service resources, that is, manual review as an example, referring to fig. 3, for a frame diagram of a video review system in the embodiment of the present application, as shown in fig. 3, a set of complete video review system frames is provided, which includes a video uploading process, a preliminary review and scheduling sorting process after successful uploading, a video scheduling and distributing process, a manual review process, and a video distributing process, and the main functions of each service module in the video review system are described first below, specifically as follows:
a. a video content producing end 10 and a video content consuming end 20.
(1) The video content producer 10, which may be a content producer such as PGC, UGC, MCN, pufc, etc., provides local or shot video mainly through a mobile terminal or a backend Application Programming Interface (API) system, which are main content sources of final video distribution, specifically: the system is used for communicating with the uplink and downlink content interface service 30, acquiring an interface address of an uploading server, and uploading a local video or a shot video, wherein matched music, a filter template, a beautifying function of the video and the like can be selected in the process of shooting the video.
(2) The video content consuming end 20 is mainly configured to communicate with the uplink and downlink content interface service 30 to obtain index information of an accessed video, and may communicate with a video content storage server (not shown in fig. 3), download a corresponding video, and play and watch the video through a local player, and the video content consuming end 20 generally browses the video in a Feeds streaming manner, so that a low-quality video has a great influence on user experience, and may also influence final duration and user stickiness.
Further, in this embodiment of the application, the video content producing end 10 and the video content consuming end 20 may also report information such as behavior data, pause, loading time, playing click times, and the like played by the user in the uploading and downloading processes to the server.
b. An uplink and downlink content interface service 30.
The method is mainly used for: (1) The video content storage server directly communicates with the video content production end 10, and the video submitted by the front end enters the server end through the uplink and downlink content interface service 30, and is stored in the video content storage server, wherein the video submitted by the front end generally comprises: the title, the publisher, the abstract, the video cover, the publishing time, etc. of the video content are not limited in the embodiments of the present application.
(2) Meta-information of the video, such as video file size, video cover link, bit rate, file format, title, release time, author, etc., is written to the video content meta-information and database 40.
(3) The uploaded video is submitted to the video content warehousing service 50 for subsequent video content processing and streaming.
c. Video content meta information and database 40.
In the embodiment of the present application, the video content meta-information and database 40 is a core database of a video, and is mainly used for: the meta-information of all the videos uploaded by the video content producing end 10 is stored, and besides the meta-information of the videos, the tags of the videos in the manual review process, such as tag information of the tags, can be stored.
Specifically, after the uplink and downlink content interface service 30 stores the uploaded video in the video content storage server, the video content storage server performs a standard transcoding operation on the video content, and asynchronously returns meta-information after transcoding is completed, for example, the meta-information includes file size, bit rate, specification, and captured cover book, and the meta-information is stored in the video content meta-information and database 40.
In addition, the video content meta-information and the meta-information in the database 40 are read in the manual review process, and the review result and the marking information of the manual review are also returned to the video content meta-information and the database 40 for storage, and the review result of the manual review is also an important basis for subsequently measuring the efficiency of the machine algorithm filter model.
d. A video content warehousing service 50.
The method is mainly used for: the scheduling service 30 receives the videos entering the database through the uplink and downlink content interface service 30, acquires meta-information of the videos from the video content meta-information and the database 40, communicates with the scheduling and sorting service 100, acquires scheduling and sorting results, i.e., scheduling priority sorting, and simultaneously dispatches and distributes corresponding videos to corresponding video content experiment buckets according to the configuration of the content scheduling bucket service 110.
e. The file system 60 is downloaded.
The method is mainly used for: the original video is downloaded and retrieved from the video content meta-information and database 40 and video content storage servers and also the speed and progress of the download can be controlled, typically a set of parallel servers, consisting of related task scheduling and distribution clusters.
f. A framing service 70.
The method is mainly used for: the video downloaded by the downloaded file system 60 is subjected to primary processing of video characteristic information, that is, key frames are extracted from the video to serve as the basis of machine processing algorithms for subsequent evaluation of definition, video cover aesthetic measure, video content understanding and the like.
In the embodiment of the present application, a possible implementation manner is provided, in which a variable-length frame extraction strategy is adopted to determine key frames in a video, for example, a scene switching frame with obvious brightness change is used, and based on the key frames, frame extraction and frame supplement are performed at equal intervals in front and back, and one video may have multiple key frames.
g. The feature extraction construction service 80.
The method is mainly used for: according to the scheduling and sequencing factor of the video cover definition level, a key frame of a video is used as input, namely the key frame is used as a video cover, the characteristics of the video are constructed by adopting a multi-mode method, and various factors for measuring the content quality of the video, such as the cover, the resolution, the code rate, whether the video is stuck, whether black edges and screens exist, whether the video is spliced, whether the video is virtualized and the like, are obtained.
h. A machine algorithm filtering service 90.
The method is mainly used for: a machine processing model is built, and according to the picture content of the video, the video to be checked, the picture content of which accords with the forbidden content condition, is filtered, namely, the video is preliminarily checked, so that the harmful picture content, such as serious bloody smell, terrorism, pornography and the like, in a part of the picture content can be filtered, and the filtered video does not participate in subsequent scheduling sequencing and manual checking.
Moreover, for the video which is not filtered out, different characteristics of the video can be marked, for example, the video definition condition, whether the video is a hot video, the degree of a title party, whether the video contains the content of a boundary ball, whether the video is excessively cut, whether the theme is not prominent, sudden stop and the like are marked according to different degrees, and the video can be used as a factor of scheduling sequencing to participate in subsequent scheduling sequencing.
Further, the machine algorithm filtering service 90 may also communicate with the statistics service 130, and read the review result of the manual review, for example, the reason of rejecting or passing the video, so as to update the preliminary review algorithm of the machine algorithm filtering service 90 as the basis for the preliminary review filtering adjustment.
i. A scheduling ordering service 100.
In the embodiment of the present application, the scheduling ordering service 100 is a service of a comparison core, and is mainly used for: and aiming at each filtered video to be audited, carrying out scheduling priority ranking according to at least one scheduling ranking factor of the video to be audited.
Wherein, at least one scheduling ordering factor comprises one or any combination of the following: the video account quality, the video hotspot level, and the video cover definition level are the same as those described in the foregoing embodiments, and are not described again here.
Moreover, the scheduling and sorting service 100 may further cooperate with the content scheduling sub-bucket service 110, and according to different audit task scheduling policies, the video audit tasks sorted based on different scheduling and sorting factors may be placed into corresponding video content experiment buckets, for example, all the video content experiment buckets that add video account quality factors, no. 1 video content experiment buckets that add video cover definition level factors, no. 2 video content experiment buckets that add video cover definition level factors, and the like.
j. A content scheduling bucketing service 110 and a video content experiment bucket.
The content scheduling bucketing service 110 is mainly used for: based on the scheduling sorting factors and the auditing task order dispatching strategy, the flow of the manual auditing receipt is segmented and configured to quantitatively measure the independent scheduling effect of each scheduling sorting factor, the scheduling effect of each independent scheduling sorting factor can be compared with the attemperation of an independent experiment bucket, and actual data can be obtained through the statistical service 130.
In order to ensure statistical significance of the experiment, the distribution of audit flow of one video content experiment barrel is usually in the order of thousands to 1W, specifically, the flow division of the video content experiment barrel can be set according to practical experience, for example, the amount of audited video is 80W per day, the video content experiment barrel is split into 100 groups, namely 100 groups of video content experiment barrels, each group of 8000 audit tasks corresponds to 400 audit receipts, namely each audit receipt comprises 20 audit tasks, so that auditors can receive the audit tasks from 100 video content experiment barrels in a random manner.
k. A video review waybill service 120.
The method is mainly used for: (1) Reading the meta-information of the video content and the meta-information of the video in the database 40 is a system developed based on a web database, which is a complex business, and can manually perform a round of preliminary filtering on the characteristics of whether the video content relates to pornography and gambling.
(2) On the basis of manual preliminary filtering, the video can be classified and labeled or confirmed, and due to the fact that machine learning of the video content is not completely mature and possibly inaccurate, secondary confirmation and labeling are needed through man-machine cooperation, and accuracy and efficiency of labeling of the video are improved.
In addition, the video audit receipt service 120 is further configured to receive traffic allocation of the content scheduling bucket service 110, and finally, an audit task of which video content experiment bucket is actually obtained is determined by the content scheduling bucket service 110 according to a rule, for example, if the number 1 bucket is added to priority ranking scheduling of video account quality, the audit task of the number 1 bucket receipt is marked specifically.
(3) And reporting the detailed flow of the audits of the sources and the audits results of the audits, the audits starting and ending time and the like in the manual auditing process to the statistical server.
l, a statistics server 130.
The method is mainly used for: and receiving the report of the consumption running water of the video auditing claim receipt service 120 and the video content consumption end 20, performing statistical mining and analysis on the reported running water for the machine algorithm filtering system 90 to read, and simultaneously providing monitoring and analysis of the content starting rate and the content auditing backlog time delay by the scheduling effect.
m, a video distribution content library 140, and a content distribution outlet service 150.
The video distribution Content library 140 is usually hierarchical, and only part of the video distribution Content library is cached near the user side, the original video Content is from a video Content storage server, and usually a group of video Content storage servers with a wide distribution range and accessed nearby near the user, and a Content Distribution Network (CDN) acceleration server is usually arranged around the video distribution Content library for distributed cache acceleration, and the video uploaded by the Content producer can be stored through the uplink and downlink Content interface service 30. After obtaining the video index information, the video content consuming end 20 may also directly access the video content storage server corresponding to the video distribution content library 140 to download the corresponding video.
The video distribution content library 140 may distribute the video to the video content consumers 20 through the content distribution outlet service 150 and may also perform personalized recall and ranking according to the user's image characteristics and the content characteristic information of the video.
For example, the corresponding video may be distributed according to a video request of the video content consuming end 20, or the video may be actively recommended through a recommendation algorithm based on the portrait features of the user, for example, the recommendation algorithm is collaborative recommendation, matrix decomposition, a supervised learning algorithm model, a deep learning-based recommendation model, and the like, which is not limited in the embodiment of the present application.
Therefore, in the embodiment of the application, a video auditing system framework is provided, videos uploaded by a video content production end are uploaded and published and then stored in a video content meta-information and database, then a machine algorithm filtering process and a manual auditing filtering process of the videos are started through a video content warehousing service, the videos which can be distributed after being confirmed in the manual auditing process finally enter a video distribution content library, then the videos in the video distribution content library are subjected to individualized recall and sequencing according to the image characteristics of a requesting user and the content characteristic information of the videos, and finally the videos are output to a consuming user of a video content consuming end, wherein the scheduling sequencing service and the content scheduling barreling service are combined, the advantages and the disadvantages of the machine algorithm and manual auditing are combined, the scheduling priority ranking is realized through the auditing results of the manual auditing, for example, the auditing does not pass the reasons, and the multi-dimensional analysis of video auditing, the scheduling sequencing effect of a plurality of scheduling sequencing factors is realized, the scheduling priority ranking is improved through the scheduling factors, the scheduling priority ranking is realized, the scheduling priority ranking is distributed to the manual auditing and the manual auditing efficiency is improved, the high-efficiency of the manual auditing is greatly reduced, the efficiency of the manual auditing is also the manual auditing is improved, and the efficiency of the manual auditing is greatly reduced.
Based on the above embodiment, the following respectively describes each process in the video review shown in fig. 3, specifically as follows:
1. and (3) a video uploading process:
referring to fig. 4, a schematic view of a video uploading process flow in the embodiment of the present application is shown, which mainly relates to a video content production end, an uplink and downlink content access service, video content meta information, and a database in a system, and specifically includes:
step 400: and the video content production end uploads the release video to the uplink and downlink content access service.
Step 401: the uplink and downlink content interface service stores and writes the video to the video content meta-information and the database.
Therefore, the process that the user uploads the video from the video content generating terminal is achieved, and the uploaded video file and the corresponding meta information are stored.
2. And (3) after successful uploading, performing preliminary audit and scheduling sequencing:
referring to fig. 5, a schematic view of a video preliminary review and scheduling sorting process in the embodiment of the present application is shown, which mainly relates to a file downloading system, a frame extracting service, a feature extraction and construction service, a machine algorithm filtering service, a scheduling sorting service, and a statistical service in the system, and specifically includes:
step 500: the download file system downloads the video from the video content meta-information and database.
Step 501: and the frame extracting service carries out frame extracting processing on the video downloaded by the file downloading system.
Step 502: the feature extraction construction service constructs content quality feature information of the video.
Specifically, the key frames extracted by the frame extraction service are used as input to construct content quality characteristic information of the video, such as the definition level of a video cover and the like.
Step 503: the machine algorithm filtering service invokes a feature retrieval service from the feature extractions.
At this time, the machine algorithm filtering service can perform preliminary audit on the video according to the picture content, and filters out the video with the picture content meeting the forbidden content condition.
Step 504: the machine algorithm filtering service reads the manual review pipelining statistics from the statistics service.
Therefore, the manual auditing result can be effectively fed back to the system, the continuous improvement and updating of the machine algorithm filtering service are realized, the manual auditing experience can be effectively utilized and accumulated, the machine algorithm filtering service can obtain preliminary training data according to the read manual auditing running statistics, can be used for measuring the filtering effect of the machine algorithm, can determine a subsequent scheduling strategy according to the effect, updates the machine algorithm filtering, and can also be used for providing guidance significance for subsequent scheduling and sequencing services, for example, if the filtering effect of the video content machine algorithm of a certain kind of problems is not good, the subsequent scheduling and sequencing services can be handed to the manual auditing.
Step 505: and the scheduling and sequencing service acquires a preliminary auditing result from the machine algorithm filtering service.
And then the scheduling and sorting service calculates scheduling and sorting scores for the videos which are not filtered out in the preliminary auditing result according to at least one scheduling and sorting factor, and performs scheduling priority sorting according to the scheduling and sorting scores.
In this way, in the embodiment of the application, the video to be audited is primarily audited, the video to be audited with the picture content meeting the forbidden content condition is filtered by the machine algorithm filtering service, and then the video is dispatched to manual auditing directly according to the time sequence, but the dispatching priority ranking is performed according to at least one dispatching ranking factor by the dispatching ranking service, so that the dispatching priority ranking result of each filtered video to be audited is obtained and then is dispatched to manual service, the accuracy and the efficiency are improved, the overstock of some time-efficient or high-quality videos is avoided, the auditing processing efficiency of the videos is improved, and the experience of the end user in watching the videos is also improved.
3. And video scheduling and distribution process:
referring to fig. 6, a schematic view of a video scheduling and distributing process flow in the embodiment of the present application is shown, which mainly relates to a video content warehousing service, a scheduling and sorting service, and a content scheduling and bucket dividing service, and specifically includes:
step 600: the video content warehousing service receives content warehousing.
Step 601: the video content warehousing service obtains a content segmentation configuration from the content scheduling bucketing service.
Step 602: and the video content warehousing service acquires a scheduling and sequencing result from the scheduling and sequencing service.
And the scheduling sorting result is the scheduling priority sorting obtained according to at least one scheduling sorting factor.
Step 603: and the video content warehousing service distributes the video scheduling to the corresponding video content experiment buckets.
Therefore, according to the configuration of the content scheduling sub-bucket service and the scheduling and sequencing result of the scheduling and sequencing service, the video scheduling needing to be audited again is distributed to the corresponding video content experiment bucket, so that the video scheduling needing to be audited again is distributed to the manual service during the subsequent manual audit.
4. And (3) manual auditing process:
referring to fig. 7, a schematic view of a process flow of manual video review in the embodiment of the present application is shown, which mainly relates to a content scheduling sub-bucket service, a video review waybill service, and a statistical service, and specifically includes:
step 700: and initiating manual review to the video review waybill service by the auditors.
Step 701: and the video auditing receipt service acquires the flow segmentation configuration of the content scheduling barrel service.
Step 702: and the video auditing waybill service acquires an auditing task from the video content experiment bucket.
And then the obtained audit task can be sent to the corresponding auditor so that the auditor can perform the audit again, and the video can be subjected to processing such as label marking.
Step 703: and the video auditing receipt service exports the rechecked videos to a video distribution content library.
Step 704: and the video examination claim list service writes the examination result of the manual examination and the label marking result information into the video content meta-information and the database.
Step 705: and the video examination and receipt service reports the manual examination and flow information to the statistical service.
Therefore, videos in the video content experiment barrel are distributed to various auditors for manual auditing, manual auditing flow information can be recorded, manual auditing experience is accumulated, and accuracy and efficiency of overall video auditing are improved.
5. And (3) video distribution process:
fig. 8 is a schematic view of a video distribution process flow in the embodiment of the present application, which mainly relates to a video content consuming side, a content distribution export service, and a video distribution content library, and specifically includes:
step 800: and the content distribution outlet service distributes videos to the video content consumption end, and the video content consumption end acquires video information streams.
Step 801: and the content distribution export service reads the video meta information and the content characteristic information, and recalls and sorts the video meta information and the content characteristic information.
And furthermore, different videos can be distributed to corresponding users by combining the picture characteristics of the users.
Based on the foregoing embodiment, referring to fig. 9, in an embodiment of the present invention, a video auditing apparatus specifically includes:
the filtering module 90 is configured to filter, according to the picture content of each video to be audited, a video to be audited whose picture content meets the prohibited content condition;
the sorting module 91 is configured to, for each filtered video to be audited, perform scheduling priority sorting according to at least one scheduling sorting factor of the video to be audited;
and the allocating module 92 is configured to allocate re-audit resources to videos to be audited, which are sorted according to the scheduling priority level and are preferentially sorted in a high-level order.
Optionally, when performing scheduling priority ranking according to at least one scheduling ranking factor of a video to be audited, the ranking module 91 is specifically configured to:
determining the value of each scheduling sorting factor according to the characteristic information of each scheduling sorting factor in each scheduling sorting factor aiming at each filtered video to be audited;
determining a scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor;
and according to the determined scheduling sorting score, performing scheduling priority sorting on each filtered video to be audited.
Optionally, if the scheduling ordering factors include at least two scheduling ordering factors, when determining the scheduling ordering score of the video to be audited according to the value of each scheduling ordering factor, the ordering module 91 is specifically configured to: and determining the scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor and the influence weight corresponding to each scheduling sorting factor.
Optionally, the at least one scheduling ranking factor is set according to an element associated with timeliness of video review; the at least one scheduling ordering factor comprises one or any combination of the following: video account quality, video hotspot level, video cover definition level.
Optionally, if the at least one scheduling ranking factor is the quality of the video account, when the value of each scheduling ranking factor is determined according to the feature information of each scheduling ranking factor in each scheduling ranking factor, the ranking module 91 is specifically configured to:
determining a video account corresponding to a video to be audited, and acquiring the total number of uploaded videos, the auditing result of each uploaded video and the playing times of each audited video within a preset time period of the corresponding video account;
determining the auditing passing rate of the video account according to the total number of the uploaded videos and the auditing result of each uploaded video;
determining account influence of the video account according to the playing times of the videos passing the audit;
and determining the quality of the video account according to the auditing passing rate and the account influence.
Optionally, if the at least one scheduling ranking factor is a video hotspot level, when the value of each scheduling ranking factor is determined according to the feature information of each scheduling ranking factor in each scheduling ranking factor, the ranking module 91 is specifically configured to:
acquiring a label of a video to be audited;
comparing the tags of the video to be audited with the pre-configured hot spot tags according to the pre-configured hot spot tags;
and determining the video hotspot grade of the video to be audited according to the number of successfully compared and matched labels of the video to be audited and the hotspot labels.
Optionally, if the at least one scheduling ranking factor is a video cover definition level, when determining a value of each scheduling ranking factor according to feature information of each scheduling ranking factor in each scheduling ranking factor, the ranking module 91 is specifically configured to:
acquiring a video cover of a video to be audited;
and respectively obtaining the definition levels determined by taking the video covers as input parameters based on a definition recognition model, wherein the definition recognition model is used for determining the definition levels according to the video covers.
Optionally, further comprising:
an obtaining module 93, configured to obtain an audit result of the re-audit resource;
and an updating module 94, configured to update the prohibited content condition and/or update the influence weight corresponding to each scheduling ranking factor according to the review result of the re-review resource.
Based on the foregoing embodiment, an electronic device according to another exemplary embodiment is further provided in this embodiment, and in some possible embodiments, the electronic device according to this embodiment may include a memory, a processor, and a computer program stored in the memory and executable on the processor, where the processor may implement the steps of the video auditing method in any of the foregoing embodiments when executing the program.
For example, taking an electronic device as the server 200 in fig. 1 of the present invention for illustration, a processor in the electronic device is the processor 210 in the server 200, and a memory in the electronic device is the memory 220 in the server 200.
Based on the foregoing embodiments, in an embodiment of the present invention, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program, when executed by a processor, implements a video auditing method in any of the above method embodiments.
As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various modifications and variations can be made in the embodiments of the present invention without departing from the spirit or scope of the embodiments of the invention. Thus, if such modifications and variations of the embodiments of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to encompass these modifications and variations.

Claims (7)

1. A video auditing method, comprising:
filtering the video to be audited with the picture content meeting the forbidden content condition according to the picture content of each video to be audited;
aiming at each filtered video to be audited, carrying out scheduling priority ranking according to at least one scheduling ranking factor of the video to be audited, wherein the at least one scheduling ranking factor is set according to elements related to timeliness of video auditing; the at least one scheduling ordering factor comprises one or any combination of the following: video account quality, video hotspot level, video cover definition level;
allocating re-audit resources for videos to be audited with high priority according to scheduling priority ranking;
the method for scheduling the videos to be audited includes the following steps:
determining the value of each scheduling sorting factor according to the characteristic information of each scheduling sorting factor in each scheduling sorting factor aiming at each filtered video to be audited;
determining a scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor;
according to the determined scheduling sorting score, performing scheduling priority sorting on each filtered video to be audited;
if the at least one scheduling ranking factor is the quality of the video account, determining the value of each scheduling ranking factor according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, specifically comprising:
determining a video account corresponding to a video to be audited, and acquiring the total number of uploaded videos, the auditing result of each uploaded video and the playing times of each audited video within a preset time period of the corresponding video account;
determining the auditing passing rate of the video account according to the total number of the uploaded videos and the auditing result of each uploaded video;
determining account influence of the video account according to the playing times of the videos passing the audit;
determining the quality of the video account according to the auditing passing rate and the account influence;
the method further comprises the following steps: and respectively calculating scheduling sorting scores based on the single scheduling sorting factor, and respectively putting the videos to be checked, which are sorted based on different single scheduling sorting factors, into different video content experiment buckets.
2. The method according to claim 1, wherein if the scheduling ranking factors include at least two, determining a scheduling ranking score of a video to be reviewed according to a value of each scheduling ranking factor, specifically including:
and determining the scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor and the influence weight corresponding to each scheduling sorting factor.
3. The method according to claim 1 or 2, wherein if the at least one scheduling ranking factor is a video hotspot level, determining a value of each scheduling ranking factor according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, specifically comprising:
acquiring a label of a video to be audited;
comparing the tags of the video to be audited with the pre-configured hot spot tags according to the pre-configured hot spot tags;
and determining the video hotspot grade of the video to be audited according to the number of successfully compared and matched labels of the video to be audited and the hotspot labels.
4. The method according to claim 1 or 2, wherein if the at least one scheduling ranking factor is a video cover definition level, determining a value of each scheduling ranking factor according to feature information of each scheduling ranking factor in each scheduling ranking factor, specifically comprising:
acquiring a video cover of a video to be audited;
and respectively obtaining the definition levels determined by taking the video covers as input parameters based on a definition recognition model, wherein the definition recognition model is used for determining the definition levels according to the video covers.
5. The method of claim 2, further comprising:
obtaining the auditing result of the re-auditing resource;
and updating the prohibited content condition and/or updating the influence weight corresponding to each scheduling sequencing factor according to the auditing result of the re-auditing resource.
6. A video review apparatus, comprising:
the filtering module is used for filtering the videos to be audited, the picture contents of which accord with the forbidden content conditions, according to the picture contents of the videos to be audited;
the sorting module is used for sorting the scheduling priority levels of each filtered video to be audited according to at least one scheduling sorting factor of the video to be audited, wherein the at least one scheduling sorting factor is set according to elements related to the timeliness of the video audit; the at least one scheduling ordering factor comprises one or any combination of the following: video account quality, video hotspot level and video cover definition level;
the distribution module is used for distributing re-audit resources to the videos to be audited with high priority according to the scheduling priority ranking;
when scheduling priority ranking is performed according to at least one scheduling ranking factor of a video to be audited, the ranking module is specifically configured to:
determining the value of each scheduling sorting factor according to the characteristic information of each scheduling sorting factor in each scheduling sorting factor aiming at each filtered video to be audited;
determining a scheduling sorting score of the video to be audited according to the value of each scheduling sorting factor;
according to the determined scheduling sorting value, performing scheduling priority sorting on each filtered video to be audited;
if the at least one scheduling ranking factor is the quality of the video account, and when the value of each scheduling ranking factor is determined according to the characteristic information of each scheduling ranking factor in each scheduling ranking factor, the ranking module is specifically configured to:
determining a video account corresponding to a video to be audited, and acquiring the total number of uploaded videos, the audit result of each uploaded video and the playing times of each audited video of the corresponding video account in a preset time period;
determining the auditing passing rate of the video account according to the total number of the uploaded videos and the auditing result of each uploaded video;
determining account influence of the video account according to the playing times of the videos passing the audit;
determining the quality of the video account according to the auditing passing rate and the account influence;
the allocation module is further configured to: and respectively calculating scheduling sorting scores based on the single scheduling sorting factor, and respectively putting the videos to be checked, which are sorted based on different single scheduling sorting factors, into different video content experiment buckets.
7. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the steps of the method of any of claims 1-5 are implemented when the program is executed by the processor.
CN201910509756.1A 2019-06-13 2019-06-13 Video auditing method and device and electronic equipment Active CN110225373B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910509756.1A CN110225373B (en) 2019-06-13 2019-06-13 Video auditing method and device and electronic equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910509756.1A CN110225373B (en) 2019-06-13 2019-06-13 Video auditing method and device and electronic equipment

Publications (2)

Publication Number Publication Date
CN110225373A CN110225373A (en) 2019-09-10
CN110225373B true CN110225373B (en) 2023-01-24

Family

ID=67816809

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910509756.1A Active CN110225373B (en) 2019-06-13 2019-06-13 Video auditing method and device and electronic equipment

Country Status (1)

Country Link
CN (1) CN110225373B (en)

Families Citing this family (21)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110493649B (en) * 2019-09-12 2021-08-20 重庆市群众艺术馆 Cultural center digital resource processing method based on public satisfaction
CN110781384B (en) * 2019-10-09 2021-06-15 腾讯科技（深圳）有限公司 Priority-based content recommendation method, device, equipment and medium
CN110944217A (en) * 2019-12-12 2020-03-31 北京达佳互联信息技术有限公司 Live broadcast information processing method and device, server and storage medium
CN111125435B (en) * 2019-12-17 2023-08-11 北京百度网讯科技有限公司 Video tag determination method and device and computer equipment
CN113207013B (en) * 2020-02-03 2023-11-17 腾讯科技（深圳）有限公司 Multimedia data release management method, device, equipment and storage medium
CN111695422B (en) * 2020-05-06 2023-08-18 Oppo（重庆）智能科技有限公司 Video tag acquisition method and device, storage medium and server
CN112749608B (en) * 2020-06-08 2023-10-17 腾讯科技（深圳）有限公司 Video auditing method, device, computer equipment and storage medium
CN111901639A (en) * 2020-07-31 2020-11-06 上海博泰悦臻电子设备制造有限公司 Multimedia video uploading method, device and system, storage medium and platform
CN112183951B (en) * 2020-09-08 2022-09-02 支付宝(杭州)信息技术有限公司 Supervision processing method and device based on audit quality evaluation
CN112153378B (en) * 2020-09-27 2023-05-26 福建天晴数码有限公司 Method and system for testing video auditing capability
CN112131428A (en) * 2020-09-30 2020-12-25 北京一点网聚科技有限公司 Video screening method and device
CN112203122B (en) * 2020-10-10 2024-01-26 腾讯科技（深圳）有限公司 Similar video processing method and device based on artificial intelligence and electronic equipment
CN112541390B (en) * 2020-10-30 2023-04-25 四川天翼网络股份有限公司 Frame extraction dynamic scheduling method and system for examination video violation analysis
CN112364216A (en) * 2020-11-23 2021-02-12 上海竞信网络科技有限公司 Edge node content auditing and filtering system and method
CN112579771A (en) * 2020-12-08 2021-03-30 腾讯科技（深圳）有限公司 Content title detection method and device
CN112689200B (en) * 2020-12-15 2022-11-11 万兴科技集团股份有限公司 Video editing method, electronic device and storage medium
CN113010739B (en) * 2021-03-18 2024-01-26 北京奇艺世纪科技有限公司 Video tag auditing method and device and electronic equipment
CN113115107B (en) * 2021-04-15 2021-12-28 深圳鸿祥源科技有限公司 Handheld video acquisition terminal system based on 5G network
CN114710692B (en) * 2022-03-22 2024-03-01 上海哔哩哔哩科技有限公司 Multimedia file processing method and device
CN115052173B (en) * 2022-06-07 2023-08-01 北京胜视京基数字文化产业发展有限公司 Content analysis method and system for network short video
CN114996165B (en) * 2022-08-01 2022-12-13 飞狐信息技术（天津）有限公司 Business data auditing method and device, storage medium and electronic equipment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101035281A (en) * 2007-04-19 2007-09-12 鲍东山 Classified content auditing system
CN101751805A (en) * 2008-12-01 2010-06-23 陈建 Video driving test practice system
US9066122B1 (en) * 2013-03-08 2015-06-23 Google Inc. Serving video content segments
WO2017166477A1 (en) * 2016-03-28 2017-10-05 乐视控股（北京）有限公司 Method and device for obtaining video information
CN109495766A (en) * 2018-11-27 2019-03-19 广州市百果园信息技术有限公司 A kind of method, apparatus, equipment and the storage medium of video audit

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8364540B2 (en) * 2005-09-14 2013-01-29 Jumptap, Inc. Contextual targeting of content using a monetization platform
CN101739416A (en) * 2008-11-04 2010-06-16 未序网络科技（上海）有限公司 Method for sequencing multi-index comprehensive weight video
US20160041998A1 (en) * 2014-08-05 2016-02-11 NFL Enterprises LLC Apparatus and Methods for Personalized Video Delivery
CN105893561A (en) * 2016-03-31 2016-08-24 北京奇艺世纪科技有限公司 Ordering method and device
CN106447239B (en) * 2016-11-21 2020-09-29 北京字节跳动科技有限公司 Data release auditing method and device
CN108170813A (en) * 2017-12-29 2018-06-15 智搜天机（北京）信息技术有限公司 A kind of method and its system of full media content intelligent checks
CN108228911A (en) * 2018-02-11 2018-06-29 北京搜狐新媒体信息技术有限公司 The computational methods and device of a kind of similar video
CN109495783A (en) * 2018-11-02 2019-03-19 平安科技（深圳）有限公司 Video reviewing method, device, electronic equipment and medium

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101035281A (en) * 2007-04-19 2007-09-12 鲍东山 Classified content auditing system
CN101751805A (en) * 2008-12-01 2010-06-23 陈建 Video driving test practice system
US9066122B1 (en) * 2013-03-08 2015-06-23 Google Inc. Serving video content segments
WO2017166477A1 (en) * 2016-03-28 2017-10-05 乐视控股（北京）有限公司 Method and device for obtaining video information
CN109495766A (en) * 2018-11-27 2019-03-19 广州市百果园信息技术有限公司 A kind of method, apparatus, equipment and the storage medium of video audit

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
《Performance Comparisons of Subjective Quality Assessment Methods for Video》;Toshiko TOMINAGA;《IEICE Transactions on Communications》;20140101;全文 *
《基于深度学习的台标检测在网络视频审核中的应用》;刘琨;《无线互联科技》;20180815(第15期);全文 *

Also Published As

Publication number Publication date
CN110225373A (en) 2019-09-10

Similar Documents

Publication Publication Date Title
CN110225373B (en) Video auditing method and device and electronic equipment
CN110399550B (en) Information recommendation method and device
CN112749608B (en) Video auditing method, device, computer equipment and storage medium
US11166076B2 (en) Intelligent viewer sentiment predictor for digital media content streams
CN108874832B (en) Target comment determination method and device
CN105431834A (en) Creating personalized and continuous playlists for a content sharing platform based on user history
WO2022252363A1 (en) Data processing method, computer device and readable storage medium
CN111310041B (en) Image-text publishing method, model training method and device and storage medium
CN112153426A (en) Content account management method and device, computer equipment and storage medium
CN106851349A (en) Based on magnanimity across the live recommendation method for shielding viewing behavior data
CN112231563B (en) Content recommendation method, device and storage medium
CN111522724B (en) Method and device for determining abnormal account number, server and storage medium
US20230004832A1 (en) Methods, Systems, And Apparatuses For Improved Content Recommendations
CN112231497A (en) Information classification method and device, storage medium and electronic equipment
CN112202849A (en) Content distribution method, content distribution device, electronic equipment and computer-readable storage medium
CN112989074A (en) Multimedia information recommendation method and device, electronic equipment and storage medium
CN113626624B (en) Resource identification method and related device
CN111191054B (en) Media data recommendation method and device
CN111460267B (en) Object identification method, device and system
US20230316106A1 (en) Method and apparatus for training content recommendation model, device, and storage medium
US11962857B2 (en) Methods, systems, and apparatuses for content recommendations based on user activity
US20230007344A1 (en) Methods, Systems, And Apparatuses For User Engagement Analysis
CN111143688B (en) Evaluation method and system based on mobile news client
CN114363660A (en) Video collection determining method and device, electronic equipment and storage medium
CN114996435A (en) Information recommendation method, device, equipment and storage medium based on artificial intelligence

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant