CN113766256A - Live broadcast wind control method and device - Google Patents

Live broadcast wind control method and device Download PDF

Info

Publication number
CN113766256A
CN113766256A CN202110181263.7A CN202110181263A CN113766256A CN 113766256 A CN113766256 A CN 113766256A CN 202110181263 A CN202110181263 A CN 202110181263A CN 113766256 A CN113766256 A CN 113766256A
Authority
CN
China
Prior art keywords
user
data
risk
live
live broadcast
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.)
Pending
Application number
CN202110181263.7A
Other languages
Chinese (zh)
Inventor
张学理
肖翔
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.)
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology Co Ltd
Original Assignee
Beijing Jingdong Century Trading Co Ltd
Beijing Wodong Tianjun Information Technology 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 Beijing Jingdong Century Trading Co Ltd, Beijing Wodong Tianjun Information Technology Co Ltd filed Critical Beijing Jingdong Century Trading Co Ltd
Priority to CN202110181263.7A priority Critical patent/CN113766256A/en
Publication of CN113766256A publication Critical patent/CN113766256A/en
Pending legal-status Critical Current

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
    • H04N21/21—Server components or server architectures
    • H04N21/218—Source of audio or video content, e.g. local disk arrays
    • H04N21/2187—Live feed
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • H04N21/24—Monitoring of processes or resources, e.g. monitoring of server load, available bandwidth, upstream 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]
    • H04N21/20—Servers specifically adapted for the distribution of content, e.g. VOD servers; Operations thereof
    • H04N21/25—Management operations performed by the server for facilitating the content distribution or administrating data related to end-users or client devices, e.g. end-user or client device authentication, learning user preferences for recommending movies
    • H04N21/258—Client or end-user data management, e.g. managing client capabilities, user preferences or demographics, processing of multiple end-users preferences to derive collaborative data
    • H04N21/25866—Management of end-user data
    • 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
    • H04N21/442—Monitoring of processes or resources, e.g. detecting the failure of a recording device, monitoring the downstream bandwidth, the number of times a movie has been viewed, the storage space available from the internal hard disk
    • 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
    • H04N21/4508—Management of client data or end-user data

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Molecular Biology (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • Computer Graphics (AREA)
  • Two-Way Televisions, Distribution Of Moving Picture Or The Like (AREA)

Abstract

The invention discloses a live broadcast wind control method and device, and relates to the technical field of computers. The specific implementation mode of the method comprises the following steps: monitoring live broadcast data in real time; determining a risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data; and when the risk value meets a preset risk condition, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy. According to the embodiment, the risk can be accurately identified and controlled, the detection and control measures are adjusted based on real-time data, the risk detection and control capacity is improved, the labor cost is reduced, the illegal content is prevented from being missed, and the normal operation of the platform is guaranteed.

Description

Live broadcast wind control method and device
Technical Field
The invention relates to the technical field of computers, in particular to a live broadcast wind control method and device.
Background
Live broadcast is a novel interactive social mode, and with the development of internet technology, users, information propagation speed, commercial value and the like of the live broadcast all show explosive growth trends.
The live broadcast risk control in the prior art generally comprises a main broadcast air control and an audience air control, wherein the main broadcast air control obtains risk contents in a mode of spot check of a platform administrator or audience report in a live broadcast room and processes the risk contents; and the audience wind control obtains the risk content in a real-time auditing mode by a live broadcast room manager, and processes the risk content.
Because the manual mode of platform administrator spot check, audience report and administrator real-time check needs to consume a large amount of manpower and material resources, the live broadcast related data is increased in an explosive manner, and the peak value of the interactive message is even up to over 1000 ten thousand pieces/s. Therefore, the risk control efficiency of live broadcast in the manual mode is low, the auditing cost is too high, the missing rate of illegal contents is extremely high due to the large magnitude difference between manual data and live broadcast data, and the platform is further likely to have the risk of off-shelf.
Disclosure of Invention
In view of this, embodiments of the present invention provide a live broadcast air control method and apparatus, which can implement accurate risk identification and control, adjust detection and control measures based on real-time data, improve risk detection and control capabilities, reduce labor cost, prevent missed detection of illegal contents, and ensure normal operation of a platform.
To achieve the above object, according to an aspect of an embodiment of the present invention, there is provided a live broadcast wind control method, including:
monitoring live broadcast data in real time;
determining a risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data;
and when the risk value meets a preset risk condition, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy.
Optionally, the live data comprises any one or more of: the method comprises the steps of interactive information sent by a first user, audio streams pushed to the first user, video streams pushed to the first user and tasks pushed to the first user and related to virtual resources.
Optionally, when the live data comprises the interactive message, the audio stream and/or the video stream, the user data comprises a target user representation; determining a risk value of the live data according to the live data and user data corresponding to a source of the live data includes:
and determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait.
Optionally, when the live data includes the interactive message, the user data includes a first user representation of a first user who generated the interactive message, the target user representation being a first user representation of the first user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
and determining whether sensitive words exist in the interactive message or not, if so, filtering the sensitive words, taking the filtered interactive message and the first user portrait as the input of a first risk prediction model, and taking the output of the first risk prediction model as a risk value corresponding to the live broadcast data.
Optionally, the preset risk condition comprises a first risk threshold; when the risk value meets a preset risk condition, the method comprises the following steps:
determining that the risk value satisfies the preset risk condition when the risk value is greater than the first risk threshold; when the risk value is not greater than the first risk threshold, displaying the filtered interactive message;
the wind control processing is carried out on the live broadcast data, and the method comprises the following steps:
and intercepting the interactive message.
Optionally, when the live data comprises the audio stream, the user data comprises a second user representation of a second user who produced the audio stream, the target user representation being a second user representation of the second user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
and converting the audio stream into a text, taking the text and the second user portrait as the input of a second risk prediction model, and taking the output of the second risk prediction model as a risk value corresponding to the live broadcast data.
Optionally, when the live data comprises the video stream, the user data comprises a second user representation of a second user who generated the video stream, the target user representation being a second user representation of the second user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
performing frame extraction on the video stream to obtain a plurality of live broadcast images corresponding to the video stream;
and taking the live broadcast image and the second user portrait as the input of a third risk prediction model, and taking the output of the third risk prediction model as a risk value corresponding to the live broadcast data.
Optionally, when the live data further includes the interactive message, the framing the video stream includes:
and determining the target frame extraction frequency according to the frequency of the interactive messages and the preset frame extraction frequency, and extracting frames of the video stream according to the target frame extraction frequency.
Optionally, the determining the target decimation rate includes:
acquiring negative feedback data of a first user for the video stream;
and determining the target frame extraction frequency according to the frequency of the negative feedback data, the frequency of the interactive messages and a preset frame extraction frequency, wherein the target frame extraction frequency is positively correlated with the frequency of the negative feedback data.
Optionally, when the live data includes the task related to the virtual resource, the user data includes a first user representation of the first user; determining a risk value of the user data according to the live data and the user data corresponding to the source of the live data includes:
acquiring real-time behavior data of the first user for the task;
and determining a risk value corresponding to the first user according to the real-time behavior data and the first user portrait.
Optionally, the wind control strategy comprises: the corresponding relation between the multiple risk levels and the multiple wind control measures, wherein the wind control measures comprise any one or more of the following measures: interrupting the live broadcast data, reducing the account number level indicated by the user data, intercepting the interactive message, adding the account number indicated by the user data into a blacklist, canceling the account number indicated by the user data, rejecting the request for executing the task, and intercepting behavior data aiming at the task.
Optionally, the performing, according to a preset wind control policy, wind control processing on the live data and/or the user data includes:
determining a target risk level corresponding to the risk value;
and executing the target risk level and the wind control measures corresponding to other risk levels lower than the target risk level to realize the wind control processing.
According to still another aspect of the embodiments of the present invention, there is provided a live broadcast wind control device, including:
the monitoring module is used for monitoring live broadcast data in real time;
the risk calculation module is used for determining a risk value of the live broadcast data and/or the user data according to the live broadcast data and the user data corresponding to the source of the live broadcast data;
and the risk processing module is used for carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy when the risk value meets a preset risk condition.
According to another aspect of the embodiments of the present invention, there is provided a live broadcast wind control electronic device, including:
one or more processors;
a storage device for storing one or more programs,
when the one or more programs are executed by the one or more processors, the one or more processors are enabled to realize the live broadcast wind control method provided by the invention.
According to a further aspect of the embodiments of the present invention, there is provided a computer readable medium, on which a computer program is stored, which when executed by a processor implements a live broadcast wind control method provided by the present invention.
One embodiment of the above invention has the following advantages or benefits: because the automatic live broadcast air control system based on the rules and the models is adopted, different air control measures are respectively adopted for audio, video, bullet screen messages, marketing activities and the like, and the method comprises the following steps: determining the risk values of the audio and the video by combining the text converted by the audio and the image converted by the video and the barrage message through a risk prediction model based on a neural network algorithm and a user portrait and carrying out account number and content wind control processing; the method comprises the steps of setting an account number and a rule for filtering flow abnormity, determining a user risk value and carrying out account number wind control processing, so that the technical means that a manual auditing mode is low in efficiency and overhigh in auditing cost, the missing rate of illegal contents is high, the technical problem that the platform has off-shelf risks is possibly caused, accurate identification and management and control of the risks can be achieved, and the detection and management and control measures are adjusted based on real-time data, so that the risk detection and management and control capacity is improved, the labor cost is reduced, the missing detection of the illegal contents is prevented, and the technical effect of normal operation of the platform is ensured.
Further effects of the above-mentioned non-conventional alternatives will be described below in connection with the embodiments.
Drawings
The drawings are included to provide a better understanding of the invention and are not to be construed as unduly limiting the invention. Wherein:
fig. 1 is an exemplary system architecture diagram of a live air control method or live air control device suitable for application to embodiments of the present invention;
fig. 2 is a schematic diagram of a main flow of a live broadcast wind control method according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a detailed flow of a live broadcast wind control method according to an embodiment of the present invention;
fig. 4 is a schematic diagram of the main modules of a live wind control device according to an embodiment of the present invention;
fig. 5 is a schematic diagram of a live air control system according to an embodiment of the present invention;
fig. 6 is a schematic block diagram of a computer system suitable for use in implementing a terminal device or server of an embodiment of the invention.
Detailed Description
Exemplary embodiments of the present invention are described below with reference to the accompanying drawings, in which various details of embodiments of the invention are included to assist understanding, and which are to be considered as merely exemplary. Accordingly, those of ordinary skill in the art will recognize that various changes and modifications of the embodiments described herein can be made without departing from the scope and spirit of the invention. Also, descriptions of well-known functions and constructions are omitted in the following description for clarity and conciseness.
The neural network algorithm comprises the following steps: an Artificial Neural Network (ANN) is formed by connecting a plurality of neurons with adjustable connection weights, and a model established based on an ANN algorithm has the characteristics of large-scale parallel processing, distributed information storage, good self-organizing and self-learning capabilities and the like.
BP algorithm: the Back Propagation algorithm, namely an error Back Propagation algorithm, is a supervised learning algorithm in an artificial neural network, can approach any function theoretically, is composed of a basic structure and a nonlinear change unit, and has strong nonlinear mapping capability; and parameters such as the number of middle layers of the network, the number of processing units of each layer, the learning coefficient of the network and the like can be set according to specific conditions, so that the method has great flexibility and has wide application prospects in many fields such as optimization, signal processing and pattern recognition, intelligent control, fault diagnosis and the like.
Fig. 1 is a diagram illustrating an exemplary system architecture of a live broadcast wind control method or a live broadcast wind control device according to an embodiment of the present invention, and as shown in fig. 1, the exemplary system architecture of the live broadcast wind control method or the live broadcast wind control device according to the embodiment of the present invention includes:
as shown in fig. 1, the system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. The network 104 serves as a medium for providing communication links between the terminal devices 101, 102, 103 and the server 105. Network 104 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The user may use the terminal devices 101, 102, 103 to interact with the server 105 via the network 104 to receive or send messages or the like. Various communication client applications, such as a live application, a shopping application, a web browser application, a search application, an instant messaging tool, a mailbox client, social platform software, and the like, may be installed on the terminal devices 101, 102, 103.
The terminal devices 101, 102, 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop portable computers, desktop computers, and the like.
The server 105 may be a server that provides various services, such as a background management server that supports live websites browsed by users using the terminal apparatuses 101, 102, and 103. The backend management server may analyze and perform other processing on the received data such as the barrage message request, and feed back a processing result (for example, allowing or rejecting sending of the barrage message) to the terminal devices 101, 102, and 103.
It should be noted that the live broadcast wind control method provided by the embodiment of the present invention is generally executed by the server 105, and accordingly, the live broadcast wind control device is generally disposed in the server 105.
It should be understood that the number of terminal devices, networks, and servers in fig. 1 is merely illustrative. There may be any number of terminal devices, networks, and servers, as desired for implementation.
Fig. 2 is a schematic diagram of a main flow of a live broadcast wind control method according to an embodiment of the present invention, and as shown in fig. 2, the live broadcast wind control method of the present invention includes:
step S201, monitoring live data in real time.
Live broadcast mode is various, for example live broadcast area goods, the live broadcast of recreation, live broadcast lottery draw, live broadcast release meeting etc. and the area of involving is extensive, so live broadcast's popularization of popularization can bring huge economic potential for the platform. However, the rapid development of live broadcasting also brings certain operation risks to the platform, and especially for live broadcasting streams with large audience numbers, the propagation speed of violation information is faster than that of the traditional information propagation mode, and the generated harm and risk are also large. In order to accurately identify and manage risks, the risk value of live broadcast data and/or user data is determined in time and subjected to wind control in a real-time monitoring mode, illegal contents can be prevented from being missed and detection is avoided, and normal operation of a platform is guaranteed. The live users may include a first user and a second user, for example, the first user is a viewer and the second user is a main cast. The live broadcast data comprises live broadcast data and user data, and the live broadcast data can comprise an interactive message sent by a first user, an audio stream pushed to the first user, a video stream pushed to the first user, a task related to a virtual resource pushed to the first user and the like; the user data may include an account number, representation, etc. of the user, such as a first user representation, a second user representation. And when live broadcast wind control is carried out, live broadcast data are monitored in real time.
Step S202, according to the live broadcast data and the user data corresponding to the source of the live broadcast data, determining the risk value of the live broadcast data and/or the user data.
When the live data comprise an interactive message sent by a first user, an audio stream pushed to the first user and a video stream pushed to the first user, determining a source of the live data according to the live data monitored in real time, and determining a risk value of the live data by using a corresponding risk prediction model based on the live data and user data corresponding to the source of the live data. For example:
when the live broadcast data comprise interactive messages sent by a first user, inputting the interactive messages sent by the first user and a first user portrait into a first risk prediction model, and taking the output of the first risk prediction model as a risk value of the interactive messages sent by the first user;
when the live data comprise an audio stream pushed to the first user, inputting the audio stream pushed to the first user and the second user portrait into a second risk prediction model, and taking the output of the second risk prediction model as a risk value of the audio stream pushed to the first user;
and when the live data comprises a video stream pushed to the first user, inputting the video stream pushed to the first user and the second user portrait into a third risk prediction model, and taking the output of the third risk prediction model as a risk value of the video stream pushed to the first user.
When the live broadcast data comprises a task related to a virtual resource pushed to the first user, determining real-time behavior data of the user according to the live broadcast data monitored in real time, and determining a risk value of the first user based on the real-time behavior data of the first user and the first user figure.
And step S203, when the risk value meets a preset risk condition, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy.
After determining the risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data, judging whether the risk value meets a preset risk condition, and if so, carrying out wind control processing on the live data and/or the user data according to a preset wind control strategy. The preset risk condition may be a risk threshold, and the preset wind control policy may include a correspondence between multiple risk levels and multiple wind control measures.
If the risk value meets the preset risk threshold, according to the risk level corresponding to the risk threshold, wind control measures are taken to carry out wind control processing on the live broadcast data and/or the user data, live broadcast stream can be interrupted, the user account level can be adjusted, messages sent by users can be intercepted, and the like.
In the embodiment of the invention, live data is monitored in real time; determining a risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data; when the risk value meets a preset risk condition, the live broadcast data and/or the user data are subjected to steps of wind control processing and the like according to a preset wind control strategy, accurate risk identification and management and control can be achieved, detection and management and control measures are adjusted based on real-time data, risk detection and management and control capacity is improved, labor cost is reduced, illegal content is prevented from being missed, and normal operation of a platform is guaranteed.
Fig. 3 is a schematic diagram of a detailed flow of a live broadcast wind control method according to an embodiment of the present invention, and as shown in fig. 3, the live broadcast wind control method of the present invention includes:
step S301, monitoring live broadcast data in real time.
The number of viewers who are on line at the same time in a hot anchor is more than million, the number of accumulated viewers per day is more than 5000 thousands, the real-time transmission speed of information is unprecedentedly improved in a live broadcast mode, but illegal information such as negative content, illegal live broadcast content, yellow gambling content and related advertisements can be rapidly transmitted, and great operation risk is brought to a live broadcast operation platform.
In order to prevent the negative effect of live broadcasting, the live broadcasting needs to be subjected to real-time wind control, various data generated in the live broadcasting needs to be monitored in real time, various data generated in the live broadcasting can include live broadcasting data and user data, and different wind control measures can be taken according to different types of live broadcasting data through real-time monitoring of the live broadcasting data.
Step S302, construct a first user representation.
The live broadcast wind control method identifies whether the live broadcast data has risks or not based on the attribute data and behavior data of the historical live broadcast of audiences and the attribute data and behavior data of real-time live broadcast, so that a first user portrait of which the first user is the audiences is constructed before the risk value of the live broadcast data is determined. During the process of watching the live broadcast, the first user may generate a plurality of user data, including user attribute data and user behavior data, where the user attribute data includes: registration information (ID, etc.), login IP address, login device, and the like; the user behavior data includes: registering, logging in, browsing, sharing, ordering, commenting, collecting, reporting, abnormal behavior and other data. According to the user attribute data and the user behavior data of the first user, processing such as data cleaning is carried out on the user data, user attribute characteristics and user behavior characteristics of the processed user data are extracted, the extracted user attribute characteristics and user behavior characteristics are input into a user portrait model, a risk label of the first user is output by the user portrait model, and a risk user portrait of the first user is constructed and serves as the first user portrait.
The risk label may include: low risk, medium risk, and high risk, the first user representation may be a risk weight matrix of user data of the first user. The account with abnormal behaviors in the historical data can be marked through the user portrait, if the abnormal behaviors occur in the history, the corresponding user portrait shows that the account possibly has high risks, and the account with a high-risk label can be processed in advance.
Illustratively, the user image model may be a neural network model based on a neural network algorithm, and the input may be historical data of the first user, including registration time, registration information, registration time, registration IP address, registration equipment, ordering, and abnormal behavior data; the abnormal behavior may include: machine registration, machine login, script operation, etc. Further, the neural network algorithm may be a BP algorithm.
Step S303, construct a second user representation.
The live broadcast wind control method identifies whether the live broadcast data has risks or not based on the attribute data and behavior data of the anchor historical live broadcast and the attribute data and behavior data of the real-time live broadcast, so that a second user portrait of which the second user is the anchor is constructed before the risk value of the live broadcast data is determined.
During the process of initiating the live broadcast, the second user may generate a plurality of user data, including user attribute data and user behavior data, where the user attribute data includes: registration information, login information, the number of fans, the type of anchor and the like; the user behavior data includes: registration, login, interaction, reporting, abnormal behavior and the like. And inputting the user attribute data and the user behavior data into the user portrait model according to the user attribute data and the user behavior data of the second user, outputting a risk label of the second user, and constructing a risk user portrait of the second user as the second user portrait.
The risk label may include: low risk, medium risk and high risk, the second user representation may be a risk weight matrix of the user data of the second user.
Illustratively, the user image model may be a neural network model based on a neural network algorithm, and the input may be historical data of the second user, including registration time, registration information, registration time, registration IP address, registration equipment, ordering, and abnormal behavior data; the abnormal behavior may include: and (4) publishing illegal statements, showing illegal pictures, delivering illegal links, carrying out abnormal interaction of vermicelli, being reported and the like. Further, the neural network algorithm may be a BP algorithm.
Step S304, judging the type of the live data.
And in the process of monitoring the live broadcast data, judging the type of the live broadcast data according to the received live broadcast data. The source of the live data may be the anchor initiating the live event or may be the viewer watching the live event, with which the viewer may interact in real time. The live broadcast data can comprise an interactive message sent by a first user, an audio stream pushed to the first user by a second user, a video stream pushed to the first user by the second user, a task related to a virtual resource pushed to the first user and the like; wherein the first user is a viewer and the second user is a main broadcast.
Exemplarily, the type of the live data is determined according to the received live data, and if the live data is an interactive message sent by the first user, the process goes to step S305; if the live data is an audio stream pushed to the first user by the second user, turning to step S308; if the live data is a video stream pushed to the first user by the second user, go to step S311; if the live data is a task related to the virtual resource pushed to the first user, go to step S314.
Step S305, preprocessing the interactive message and the source thereof sent by the first user.
When the live broadcast data is the interactive message sent by the first user, the interactive message sent by the first user and the source of the interactive message are preprocessed, so that the wind control processing efficiency is improved, and the method comprises the following steps:
a1: judging whether the account number of the first user is a high-risk account number or not according to the first user portrait constructed in the step S302, and if so, turning to a step S316; if not, go to step A2.
A2: judging whether the account number of the first user is a blacklist account number, if so, turning to step S316; if not, go to step A3.
A3: judging whether the account login channel of the first user is abnormal, if so, turning to the step S316; if not, go to step A4.
A4: judging whether the sending of the interactive message of the first user is carried out through an automatic script forgery message protocol, if so, turning to the step S316; if not, go to step A5. The message sending of the first user can use a protocol established by both the message transceiver to carry out communication, the protocol content comprises information such as the identity and equipment of the user, the black generation account number can log in through information such as forged identity and equipment, and great risk can be generated on the live broadcast platform, therefore, by judging whether the sending of the interactive message of the first user is carried out through an automatic script forged message protocol, wind control measures can be further adopted, and the risk of the platform is reduced.
A5: judging whether sensitive words exist in the interactive message of the first user according to a preset sensitive word bank, if so, filtering the sensitive words and then turning to the step S306; if not, go to step S306.
By judging the high-risk account, the blacklist account, the login channel and the like, and identifying and filtering the sensitive words based on the sensitive word rule, the account with obvious risk characteristics can be identified in real time, and further wind control processing is performed.
Illustratively, the interactive message sent by the first user may be a bullet screen message, a behavior message, a comment message, and the like sent by the first user in real time; the action message may be, for example: the user enters the live broadcast room, the user exits the live broadcast room, and the like.
Step S306, determining the risk value of the interactive message sent by the first user.
After interactive messages sent by a first user are preprocessed, inputting the interactive messages sent by the first user and a first user portrait into a first risk prediction model, outputting risk values by the first risk prediction model, and determining the risk values of the interactive messages sent by the first user according to the output of the first risk prediction model; wherein the interactive message may be text.
Illustratively, the first risk prediction model may be a neural network model based on a neural network algorithm, the real-time interactive messages which may have risks are identified through the first risk prediction model, and effective identification may be performed on contents which may not be identified based on sensitive word rules, for example, messages of types such as deformed words, English pinyin mixed rows, and interspersed special symbols, so as to further improve the risk identification capability of the platform and prevent omission of high-risk information. Further, the neural network algorithm may be a BP algorithm.
Illustratively, the risk value may be any value from 0 to 1, and the closer the risk value is to 1, the higher the risk of the interactive message sent by the first user is represented.
Step S307, determining whether the risk value of the interactive message sent by the first user meets a preset risk condition.
The preset risk condition comprises a first risk threshold, after the risk value of the interactive message sent by the first user is determined according to the step S306, whether the risk value of the interactive message sent by the first user is larger than the first risk threshold is judged, if yes, the risk value of the interactive message sent by the first user meets the preset risk condition is judged, and the step S316 is executed; and if not, indicating that the risk value of the interactive message sent by the first user does not meet the preset risk condition, allowing the first user to send the interactive message to a second user, one or more other first users and the like.
For example, for an interactive message subjected to the sensitive word filtering processing, if the risk value of the interactive message sent by the first user does not satisfy the preset risk condition, the sensitive word may be specially processed and then sent to the interactive message.
Illustratively, the more forward tags a user portrait is, the higher the risk threshold may be.
And step S308, preprocessing the audio stream pushed to the first user by the second user.
When the live broadcast data is the audio stream pushed to the first user by the second user, the audio stream pushed to the first user by the second user is preprocessed, so that the wind control processing efficiency is improved, and the method comprises the following steps:
converting the audio stream pushed to the first user by the second user into a text;
cleaning data of the converted text, wherein the data comprises information such as repeated characters, abnormal characters (blank spaces, messy codes and the like), redundant data, invalid data and the like in the deleted text; correcting obviously recognizable error information such as wrongly written characters, wrongly punctuated marks and the like; filling missing information and the like so as to ensure the integrity and consistency of data and meet the quality requirement of the subsequent flow.
For example, the converted text may be subjected to sensitive word filtering processing, thereby improving the efficiency of the wind control processing.
Step S309, determining a risk value of the audio stream pushed by the second user to the first user.
And after preprocessing the audio stream pushed to the first user by the second user, inputting the preprocessed text and the second user portrait into a second risk prediction model, outputting a risk value by the second risk prediction model, and determining the risk value of the audio stream pushed to the first user by the second user according to the output of the second risk prediction model.
For example, the second risk prediction model may be a neural network model based on a neural network algorithm, and the second risk prediction model may be the same as the first risk prediction model, and the risk prediction model may perform similarity calculation on the text, thereby determining a risk value of the text. Further, the neural network algorithm may be a BP algorithm.
Illustratively, the risk value may be any value from 0 to 1, the closer the risk value is to 1, the higher the risk representing the audio stream pushed by the second user to the first user.
Step S310, determining whether the risk value of the audio stream pushed by the second user to the first user meets a preset risk condition.
The preset risk condition comprises a second risk threshold, after the risk value of the audio stream pushed to the first user by the second user is determined according to the step S309, whether the risk value of the audio stream pushed to the first user by the second user is larger than the second risk threshold is judged, if yes, the risk value of the audio stream pushed to the first user by the second user meets the preset risk condition is indicated, and the step S316 is carried out; if not, the risk value of the audio stream pushed to the first user by the second user does not meet the preset risk condition, and the second user is allowed to push the audio stream to the first user.
Illustratively, the more forward tags a user portrait is, the higher the risk threshold may be.
For example, in the case of gambling and pornography content, audio content often has obvious language features, such as speech teasing, induced recharging, induced grouping and the like, and by converting voice into text for recognition, possible risk information can be quickly judged and risk prompts can be given.
And step 311, preprocessing the video stream pushed to the first user by the second user.
When the live broadcast data is a video stream pushed to the first user by the second user, the video stream pushed to the first user by the second user is preprocessed, so that the wind control processing efficiency is improved, and the method comprises the following steps:
presetting a frame extraction frequency for extracting image data from a video stream as a preset frame extraction frequency; wherein the preset frame drawing frequency can be once per minute;
in the process of initiating a video stream by a second user, acquiring an interactive message sent by a first user in real time, and determining the frequency of the interactive message sent by the first user and the similarity of the interactive messages sent among a plurality of first users according to the interactive message sent by the first user;
inputting the frequency of the first user sending the interactive messages, the similarity of the interactive messages and the second user portrait into an image frame extraction model, and outputting frame extraction sample values by the image frame extraction model; the frame extracting sample value can be any value from 0 to 1, the closer the frame extracting sample value is to 1, the higher the corresponding frame extracting frequency is, the closer the frame extracting sample value is to 0, the lower the corresponding frame extracting frequency is, for example, the interactive message can be a barrage message, and if the fluctuation of the barrage quantity of the live broadcast is obvious, the higher the risk of the live broadcast is, the higher the frame extracting sample value is;
in the process of initiating a video stream by a second user, acquiring negative feedback data of the first user in real time, and adjusting the frame-extracting frequency according to the frequency of the negative feedback data of the first user; the negative feedback data can be report data and the like, the higher the frequency of the negative feedback data is, the higher the corresponding frame extraction frequency is, and the lower the frequency of the negative feedback data is, the lower the corresponding frame extraction frequency is;
on the basis of the preset frame extraction frequency, determining the target frame extraction frequency according to the frame extraction sample value and the negative feedback data of the first user; the target frame extraction frequency is positively correlated with the frequency of negative feedback data of the first user;
and performing frame extraction on the video stream pushed to the first user by the second user according to the target frame extraction frequency, and converting the extracted image data into a live image so as to meet the input requirements of a third risk prediction model, such as adjusting the resolution of the image, cutting the size of the image, converting the format of the image and the like.
By dynamic adjustment of the frame extraction video for portrait and real-time reporting data, the risk identification accuracy of violation content can be improved.
Illustratively, the image frame-decimation model includes a plurality of, for example, toC-class live image frame-decimation models and toB-class live image frame-decimation models, which are trained separately and used to determine the frame-decimation frequency of the video stream. The toC-type live broadcast refers to client live broadcast, and brand exposure is increased through live broadcast for sale; the ToB type live broadcast refers to business live broadcast, and continuous sales of enterprises are supported through live broadcast mining. toC live broadcast can be the mode of "red & star & scene + live broadcast platform", and toB can be the mode of "premium content + online customer service".
Step S312, determining a risk value of the video stream pushed by the second user to the first user.
After video streams pushed to the first user by the second user are preprocessed, the preprocessed live broadcast images and the second user portrait are input into a third risk prediction model, the third risk prediction model outputs risk values, and the risk values of the video streams pushed to the first user by the second user are determined according to the output of the third risk prediction model.
Illustratively, the third risk prediction model may be a neural network model based on a neural network algorithm, and the risk prediction model may perform similarity calculation on the image to determine a risk value of the image. Further, the neural network algorithm may be a BP algorithm.
Illustratively, the risk value may be any value from 0 to 1, and the closer the risk value is to 1, the higher the risk of representing the video stream pushed by the second user to the first user.
Step S313, determining whether the risk value of the video stream pushed by the second user to the first user meets a preset risk condition.
The preset risk condition includes a third risk threshold, after determining the risk value of the video stream pushed by the second user to the first user according to step S312, it is determined whether the risk value of the video stream pushed by the second user to the first user is greater than the third risk threshold, if so, it indicates that the risk value of the video stream pushed by the second user to the first user meets the preset risk condition, and the process goes to step S316; if not, the risk value of the video stream pushed to the first user by the second user is not met with the preset risk condition, and the second user is allowed to push the video stream to the first user.
Illustratively, the more forward tags a user portrait is, the higher the risk threshold may be.
Step S314, when the live data is a task related to a virtual resource pushed to the first user, determining a risk value of the first user.
When the live data is a task related to the virtual resource pushed to the first user, setting participation conditions which need to be met by the task related to the virtual resource pushed to the first user and the related users capable of executing the task.
Determining a risk value of a first user by judging the condition of the first user, comprising:
b1: judging whether the account number of the first user is a blacklist account number, if so, turning to step S316; if not, go to step B2.
B2: judging whether the account number of the first user is a high-risk account number or not according to the first user portrait constructed in the step S302, and if so, turning to a step S316; if not, go to step B3.
B3: judging whether the first user meets the participation condition that the task can be executed, if so, turning to the step B4; if not, go to step S316.
B4: and acquiring real-time behavior data of the first user for the task related to the virtual resource, determining a risk value corresponding to the first user according to the real-time behavior data and the first user portrait, and turning to the step S315. The real-time behavior data may be real-time traffic data, and the real-time traffic fluctuation is obvious, which indicates that the risk of the user may be higher and the corresponding risk value is also higher.
By judging the blacklist account, the high-risk account, the participation condition, the real-time behavior data and the like, the account with obvious risk characteristics can be identified in real time, and further wind control processing is performed.
Illustratively, the task related to the virtual resource pushed to the first user may be a coupon, a lottery, a rush purchase and other activities pushed to the first user in the live broadcast marketing activity, and the participation condition that the related user who can execute the task needs to meet may be a condition that the first user needs to meet the activity, the interaction frequency, the order-taking frequency and other conditions of the historical live broadcast marketing activity.
Step S315, determining whether the risk value of the first user meets a preset risk condition.
The preset risk condition includes a fourth risk threshold, after the risk value of the first user is determined according to the step S314, whether the risk value of the first user is greater than the fourth risk threshold is judged, if so, the risk value of the first user is indicated to meet the preset risk condition, and the process goes to a step S316; and if not, indicating that the risk value of the first user does not meet the preset risk condition, allowing the first user to execute the task related to the virtual resource pushed to the first user.
Illustratively, the more forward tags a user portrait is, the higher the risk threshold may be.
And step S316, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy.
The preset wind control strategy comprises corresponding relations between various risk levels and various wind control measures, and the risk levels can comprise: low risk, medium risk, and high risk, the wind control measures may include: interrupting live broadcast data, adjusting account number level indicated by user data, intercepting interactive messages, adding account numbers indicated by the user data into a blacklist, cancelling account numbers indicated by the user data, rejecting requests for executing tasks, intercepting behavior data aiming at the tasks and the like. The low risk rating may: adjusting the account number grade indicated by the user data; the intermediate risk level may: adjusting the account level indicated by the user data, intercepting the interactive messages, intercepting the behavior data aiming at the tasks and the like; the high risk rating may: interrupting live broadcast data, intercepting interactive messages, adding an account number indicated by user data into a blacklist, cancelling the account number indicated by the user data, rejecting a request for executing a task and the like.
Illustratively, if the user account is a high risk account, intercepting the interactive message; and adding the account number indicated by the user data into a blacklist, and/or logging off the account number indicated by the user data, and/or rejecting the request for executing the task.
Intercepting the interaction message if the user account is a blacklist account; and, the account indicated by the user data may be logged out.
Intercepting the interaction message if the user account logs in the channel abnormally; and, the user account level may be lowered, and/or the account indicated by the user data may be logged out.
Intercepting the interactive message if the first user sends the interactive message through an automatic script forgery message protocol; and the account number level indicated by the first user can be reduced, and/or the account number indicated by the first user is added into a blacklist, and/or the account number indicated by the first user is logged off.
If the risk value of the interactive message sent by the first user meets the preset risk condition, returning a sending refusing state to the first user; and, intercepting the interactive message. In addition, the account number level indicated by the first user can be lowered, and/or the account number indicated by the first user is added to a blacklist, and/or the account number indicated by the first user is logged off.
If the risk value of the audio stream pushed to the first user by the second user meets a preset risk condition or the risk value of the video stream pushed to the first user by the second user meets the preset risk condition, pushing a risk prompt to the second user, and if the second user triggers the preset risk condition for multiple times, interrupting live broadcast data, and/or adding an account number indicated by the second user into a blacklist, and/or cancelling the account number indicated by the second user.
If the first user can not meet the participation condition of the executable task, rejecting the request of the first user for executing the task; and adjusting the account level indicated by the first user, and/or intercepting the behavior data of the first user for the task.
If the risk value of the first user meets the preset risk condition, adjusting the account level indicated by the first user, and/or adding the account indicated by the first user into a blacklist, and/or canceling the account indicated by the first user, and/or rejecting a request of the first user for executing a task, and/or intercepting behavior data of the first user for the task.
By aiming at the automatic risk prompt of the risk content and adopting gradually enhanced wind control measures, the investment of manpower audit and inspection cost is obviously reduced, and the perfect wind control self-operation capability is demonstrated. Through the risk identification of the user account based on the user portrait, the risk account is identified in real time, the participation degree of malicious accounts is reduced, the wind control capability of the user account is improved, and the risk account can be directly shielded and processed.
Through real-time intervention on abnormal flow behaviors of marketing activities, for example, common commodity rush purchases obviously lower than the market price, the traditional marketing rules cannot identify a rush purchase script and other malicious requests, the commodity is possibly rush purchased by a black script account, normal users cannot purchase the commodity, the activity requests with abnormal flow characteristics can be processed in time, and the live risk is reduced.
Exemplary, the request to reject execution of the task may be: setting a button corresponding to the task to be not triggerable (for example, the 'buy-in' button is gray), so that the first user cannot trigger the task; intercepting behavior data for a task may be: the button corresponding to the task can be triggered, but the first user cannot obtain corresponding data after being triggered (for example, the 'purchase' button is normal, but the user does not react after being triggered).
For example, a wind control duration corresponding to each wind control measure may be set, and after a predetermined wind control duration, the corresponding wind control measure may be cancelled. For example, after a predetermined duration of time, the user account is de-sealed and deleted from the blacklist.
Exemplarily, if the risk value of the interactive message sent by the first user meets a preset risk condition, returning a sending refusal state to the first user; and, intercepting the interactive message. In addition, the account number level indicated by the first user can be lowered, and/or the account number indicated by the first user is added to a blacklist, and/or the account number indicated by the first user is logged off.
Step S317, risk data post-processing.
When the live data and/or the user data are judged to have risks, recording risk events in real time, storing corresponding risk data, updating sample data of the user portrait model based on the stored risk data, and updating the user portrait; and updating the risk prediction model according to the updated user portrait and the updated sample data of the risk prediction model, so as to realize real-time and accurate risk identification.
Illustratively, the image extraction model is updated according to the updated user portrait and the sample data of the updated image extraction model.
Compared with the traditional manual auditing mode of platform administrator spot inspection, audience reporting and administrator real-time auditing, a large amount of manpower is consumed, and because of huge data amount, the manual auditing can not cover all live broadcast data, the auditing cost is high, the number of missed inspections is large, the processing speed is low and the like, the method is easy to be maliciously damaged by automatic scripts, wool parties and the like, the normal marketing activities are interfered, unknown risks can not be faced, and the platform risk potential is very large. The risk problem in the live broadcast can be effectively solved, and the method has extremely high popularization value.
In the embodiment of the invention, live data is monitored in real time; constructing a first user representation; constructing a second user representation; judging the type of live data; preprocessing an interactive message sent by a first user and a source thereof; determining a risk value of an interactive message sent by a first user; judging whether a risk value of an interactive message sent by a first user meets a preset risk condition; preprocessing an audio stream pushed to the first user by the second user; determining a risk value of an audio stream pushed by a second user to a first user; judging whether the risk value of the audio stream pushed to the first user by the second user meets a preset risk condition or not; preprocessing a video stream pushed to a first user by a second user; determining a risk value of a video stream pushed to a first user by a second user; judging whether the risk value of the video stream pushed to the first user by the second user meets a preset risk condition or not; when the live broadcast data is a task related to virtual resources and pushed to a first user, determining a risk value of the first user; judging whether the risk value of the first user meets a preset risk condition or not; carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy; risk data post-processing and other steps can realize accurate identification and management and control of risks, and based on real-time data adjustment detection and management and control measures, risk detection and management and control capabilities are improved, labor cost is reduced, illegal contents are prevented from being missed to be detected, and normal operation of the platform is guaranteed.
Fig. 4 is a schematic diagram of main modules of a live broadcast wind control device according to an embodiment of the present invention, and as shown in fig. 4, a live broadcast wind control device 400 of the present invention includes:
and the monitoring module 401 is configured to monitor live data in real time.
The live users may include a first user and a second user, for example, the first user is a viewer and the second user is a main cast. The live broadcast data comprises live broadcast data and user data, and the live broadcast data can comprise an interactive message sent by a first user, an audio stream pushed to the first user, a video stream pushed to the first user, a task related to a virtual resource pushed to the first user and the like; the user data may include an account number, representation, etc. of the user, such as a first user representation, a second user representation. When live broadcast wind control is performed, the monitoring module 401 monitors live broadcast data in real time.
A risk calculation module 402, configured to determine a risk value of the live data and/or the user data according to the live data and the user data corresponding to a source of the live data.
When the live data includes an interactive message sent by a first user, an audio stream pushed to the first user, and a video stream pushed to the first user, a source of the live data is determined according to the live data monitored in real time by the monitoring module 401, and the risk calculation module 402 determines a risk value of the live data by using a corresponding risk prediction model based on the live data and user data corresponding to the source of the live data. For example:
when the live broadcast data comprises an interactive message sent by a first user, the risk calculation module 402 inputs the interactive message sent by the first user and a first user portrait into a first risk prediction model, and takes the output of the first risk prediction model as a risk value of the interactive message sent by the first user;
when the live data includes an audio stream pushed to the first user, the risk calculation module 402 inputs the audio stream pushed to the first user and the second user portrait into the second risk prediction model, and takes the output of the second risk prediction model as a risk value of the audio stream pushed to the first user;
when the live data includes a video stream pushed to the first user, the risk calculation module 402 inputs the video stream pushed to the first user and the second user portrait into the third risk prediction model, and takes an output of the third risk prediction model as a risk value of the video stream pushed to the first user.
When the live data includes a task related to a virtual resource pushed to the first user, the risk calculation module 402 determines real-time behavior data of the user according to the live data monitored in real time, and determines a risk value of the first user based on the real-time behavior data of the first user and the first user profile.
And a risk processing module 403, configured to perform wind control processing on the live broadcast data and/or the user data according to a preset wind control policy when the risk value meets a preset risk condition.
After the risk calculation module 402 determines the risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data, the risk processing module 403 determines whether the risk value meets a preset risk condition, and if so, the risk processing module 403 performs wind control processing on the live data and/or the user data according to a preset wind control strategy. The preset risk condition may be a risk threshold, and the preset wind control policy may include a correspondence between multiple risk levels and multiple wind control measures.
If the risk value meets the preset risk threshold, the risk processing module 403 performs wind control processing on the live broadcast data and/or the user data by taking wind control measures according to the risk level corresponding to the risk threshold, and may interrupt the live broadcast stream, adjust the user account level, intercept the message sent by the user, and the like.
In the embodiment of the invention, through the monitoring module, the risk calculation module, the risk processing module and other modules, the accurate identification and control of risks can be realized, the detection and control measures are adjusted based on real-time data, the risk detection and control capability is improved, the labor cost is reduced, the omission of illegal contents is prevented, and the normal operation of a platform is ensured.
Fig. 5 is a schematic view of a live broadcast wind control system according to an embodiment of the present invention, and as shown in fig. 5, the live broadcast wind control system of the present invention includes:
the method comprises four parts of live broadcast application, a live broadcast gateway, live broadcast air control and basic capability.
Live applications include live audio systems, live video systems, live barrage systems, live marketing systems, and the like.
The direct broadcast gateway provides a uniform direct broadcast wind control system gateway, and gateway capabilities such as request authentication, user authentication, data point burying, protocol adaptation and the like are realized.
The live broadcast wind control comprises a wind control flow control engine, a wind control model engine, a wind control rule engine and the like. The system comprises a wind control model engine, a risk prediction model, an image frame extraction model and the like, wherein the wind control model engine provides a user portrait model, the risk prediction model, the image frame extraction model and the like, the user portrait model is used for marking risks of a live user account, the risk prediction model is used for identifying risk contents of texts and images, and the image frame extraction model is used for determining frame extraction frequency of a live video; the rule engine provides real-time filtering rules of black and white lists, risk account numbers, message protocol sensitive words and the like; the process control engine is used for arranging and controlling the wind control processes of live accounts, content and the like, and can directly intercept accounts and content which do not accord with rules.
The basic service provides basic capabilities of file storage, data caching, audio-to-text conversion, video-to-image conversion and the like, and provides basic technical support for a live broadcast wind control system.
Fig. 6 is a schematic structural diagram of a computer system suitable for implementing a terminal device according to an embodiment of the present invention, and as shown in fig. 6, the computer system 600 of the terminal device according to the embodiment of the present invention includes:
a Central Processing Unit (CPU)601, which can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)602 or a program loaded from a storage section 608 into a Random Access Memory (RAM) 603. In the RAM603, various programs and data necessary for the operation of the system 600 are also stored. The CPU601, ROM602, and RAM603 are connected to each other via a bus 604. An input/output (I/O) interface 606 is also connected to bus 604.
The following components are connected to the I/O interface 606: an input portion 606 including a keyboard, a mouse, and the like; an output portion 607 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage section 608 including a hard disk and the like; and a communication section 609 including a network interface card such as a LAN card, a modem, or the like. The communication section 609 performs communication processing via a network such as the internet. The driver 610 is also connected to the I/O interface 606 as needed. A removable medium 611 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 610 as necessary, so that a computer program read out therefrom is mounted in the storage section 608 as necessary.
In particular, according to the embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 609, and/or installed from the removable medium 611. The computer program performs the above-described functions defined in the system of the present invention when executed by the Central Processing Unit (CPU) 601.
It should be noted that the computer readable medium shown in the present invention can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present invention, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In the present invention, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The modules described in the embodiments of the present invention may be implemented by software or hardware. The described modules may also be provided in a processor, which may be described as: a processor includes a monitoring module, a risk calculation module, and a risk processing module. The names of these modules do not in some cases constitute a limitation on the modules themselves, for example, the risk processing module may also be described as a "module for performing wind control processing on live data and/or user data according to a preset wind control policy".
As another aspect, the present invention also provides a computer-readable medium that may be contained in the apparatus described in the above embodiments; or may be separate and not incorporated into the device. The computer readable medium carries one or more programs which, when executed by a device, cause the device to comprise: monitoring live broadcast data in real time; determining a risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data; and when the risk value meets a preset risk condition, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy.
According to the technical scheme of the embodiment of the invention, the risk can be accurately identified and controlled, the detection and control measures are adjusted based on real-time data, the risk detection and control capability is improved, the labor cost is reduced, the illegal content is prevented from being missed, and the normal operation of the platform is ensured.
The above-described embodiments should not be construed as limiting the scope of the invention. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions can occur, depending on design requirements and other factors. Any modification, equivalent replacement, and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (15)

1. A live broadcast wind control method is characterized by comprising the following steps:
monitoring live broadcast data in real time;
determining a risk value of the live data and/or the user data according to the live data and the user data corresponding to the source of the live data;
and when the risk value meets a preset risk condition, carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy.
2. The method of claim 1,
the live data includes any one or more of: the method comprises the steps of interactive information sent by a first user, audio streams pushed to the first user, video streams pushed to the first user and tasks pushed to the first user and related to virtual resources.
3. The method of claim 2, wherein when the live data comprises the interactive message, the audio stream, and/or the video stream, the user data comprises a target user representation; determining a risk value of the live data according to the live data and user data corresponding to a source of the live data includes:
and determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait.
4. The method of claim 3, wherein when the live data comprises the interactive message, the user data comprises a first user representation of a first user who generated the interactive message, the target user representation being a first user representation of the first user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
and determining whether sensitive words exist in the interactive message or not, if so, filtering the sensitive words, taking the filtered interactive message and the first user portrait as the input of a first risk prediction model, and taking the output of the first risk prediction model as a risk value corresponding to the live broadcast data.
5. The method of claim 4, wherein the preset risk condition comprises a first risk threshold; when the risk value meets a preset risk condition, the method comprises the following steps:
determining that the risk value satisfies the preset risk condition when the risk value is greater than the first risk threshold; when the risk value is not greater than the first risk threshold, displaying the filtered interactive message;
the wind control processing is carried out on the live broadcast data, and the method comprises the following steps:
and intercepting the interactive message.
6. The method of claim 3, wherein when the live data comprises the audio stream, the user data comprises a second user representation of a second user who produced the audio stream, the target user representation being a second user representation of the second user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
and converting the audio stream into a text, taking the text and the second user portrait as the input of a second risk prediction model, and taking the output of the second risk prediction model as a risk value corresponding to the live broadcast data.
7. The method of claim 3, wherein when the live data comprises the video stream, the user data comprises a second user representation of a second user who generated the video stream, the target user representation being a second user representation of the second user; determining a risk value corresponding to the live broadcast data by using a risk prediction model according to the live broadcast data and the target user portrait, including:
performing frame extraction on the video stream to obtain a plurality of live broadcast images corresponding to the video stream;
and taking the live broadcast image and the second user portrait as the input of a third risk prediction model, and taking the output of the third risk prediction model as a risk value corresponding to the live broadcast data.
8. The method of claim 7, wherein when the live data further comprises the interactive message, the framing the video stream comprises:
and determining the target frame extraction frequency according to the frequency of the interactive messages and the preset frame extraction frequency, and extracting frames of the video stream according to the target frame extraction frequency.
9. The method of claim 8, wherein determining the target decimation rate comprises:
acquiring negative feedback data of a first user for the video stream;
and determining the target frame extraction frequency according to the frequency of the negative feedback data, the frequency of the interactive messages and a preset frame extraction frequency, wherein the target frame extraction frequency is positively correlated with the frequency of the negative feedback data.
10. The method of claim 2, wherein when the live data includes the task related to the virtual resource, the user data includes a first user representation of the first user; determining a risk value of the user data according to the live data and the user data corresponding to the source of the live data includes:
acquiring real-time behavior data of the first user for the task;
and determining a risk value corresponding to the first user according to the real-time behavior data and the first user portrait.
11. The method according to any one of claims 1 to 10,
the wind control strategy comprises the following steps: the corresponding relation between the multiple risk levels and the multiple wind control measures, wherein the wind control measures comprise any one or more of the following measures: interrupting the live broadcast data, reducing the account number level indicated by the user data, intercepting the interactive message, adding the account number indicated by the user data into a blacklist, canceling the account number indicated by the user data, rejecting the request for executing the task, and intercepting behavior data aiming at the task.
12. The method according to claim 11, wherein the performing a wind control process on the live data and/or the user data according to a preset wind control policy includes:
determining a target risk level corresponding to the risk value;
and executing the target risk level and the wind control measures corresponding to other risk levels lower than the target risk level to realize the wind control processing.
13. A live wind accuse device which characterized in that includes:
the monitoring module is used for monitoring live broadcast data in real time;
the risk calculation module is used for determining a risk value of the live broadcast data and/or the user data according to the live broadcast data and the user data corresponding to the source of the live broadcast data;
and the risk processing module is used for carrying out wind control processing on the live broadcast data and/or the user data according to a preset wind control strategy when the risk value meets a preset risk condition.
14. A live air control electronic device, comprising:
one or more processors;
a storage device for storing one or more programs,
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-12.
15. A computer-readable medium, on which a computer program is stored, which, when being executed by a processor, carries out the method according to any one of claims 1-12.
CN202110181263.7A 2021-02-09 2021-02-09 Live broadcast wind control method and device Pending CN113766256A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110181263.7A CN113766256A (en) 2021-02-09 2021-02-09 Live broadcast wind control method and device

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110181263.7A CN113766256A (en) 2021-02-09 2021-02-09 Live broadcast wind control method and device

Publications (1)

Publication Number Publication Date
CN113766256A true CN113766256A (en) 2021-12-07

Family

ID=78786618

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110181263.7A Pending CN113766256A (en) 2021-02-09 2021-02-09 Live broadcast wind control method and device

Country Status (1)

Country Link
CN (1) CN113766256A (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114205676A (en) * 2021-12-08 2022-03-18 广州方硅信息技术有限公司 Live broadcast monitoring method, device, medium and computer equipment
CN114679600A (en) * 2022-03-24 2022-06-28 上海哔哩哔哩科技有限公司 Data processing method and device
CN114913172A (en) * 2022-07-13 2022-08-16 广东电网有限责任公司佛山供电局 Method, system, equipment and medium for identifying manufacturing risk of cable middle head
CN116072123A (en) * 2023-03-06 2023-05-05 南昌航天广信科技有限责任公司 Broadcast information playing method and device, readable storage medium and electronic equipment

Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105871933A (en) * 2016-06-22 2016-08-17 腾讯科技（深圳）有限公司 Virtual asset allocation system, virtual asset allocation method and virtual asset allocation device
CN107197331A (en) * 2017-05-03 2017-09-22 北京奇艺世纪科技有限公司 A kind of method and device of real-time monitoring live content
CN108932451A (en) * 2017-05-22 2018-12-04 北京金山云网络技术有限公司 Audio-video frequency content analysis method and device
CN109766472A (en) * 2018-12-28 2019-05-17 广州华多网络科技有限公司 Signal auditing method, device, electronic equipment and storage medium
CN109831459A (en) * 2019-03-22 2019-05-31 百度在线网络技术（北京）有限公司 Method, apparatus, storage medium and the terminal device of secure access
CN109831698A (en) * 2018-12-28 2019-05-31 广州华多网络科技有限公司 Signal auditing method, device, electronic equipment and computer-readable storage medium
CN110990631A (en) * 2019-12-16 2020-04-10 腾讯科技（深圳）有限公司 Video screening method and device, electronic equipment and storage medium
CN111182314A (en) * 2018-11-12 2020-05-19 阿里巴巴集团控股有限公司 Live stream processing method and device and data processing method
CN111432227A (en) * 2020-03-27 2020-07-17 广州酷狗计算机科技有限公司 Virtual resource transfer risk determination method, device, server and storage medium
CN111711831A (en) * 2020-06-28 2020-09-25 腾讯科技（深圳）有限公司 Data processing method and device based on interactive behavior and storage medium
CN111724069A (en) * 2020-06-22 2020-09-29 百度在线网络技术（北京）有限公司 Method, apparatus, device and storage medium for processing data
CN111753520A (en) * 2020-06-02 2020-10-09 五八有限公司 Risk prediction method and device, electronic equipment and storage medium
CN111770353A (en) * 2020-06-24 2020-10-13 北京字节跳动网络技术有限公司 Live broadcast monitoring method and device, electronic equipment and storage medium
CN111770352A (en) * 2020-06-24 2020-10-13 北京字节跳动网络技术有限公司 Security detection method and device, electronic equipment and storage medium
CN112199640A (en) * 2020-09-30 2021-01-08 广州市百果园网络科技有限公司 Abnormal user auditing method and device, electronic equipment and storage medium
CN112312152A (en) * 2020-10-27 2021-02-02 浙江集享电子商务有限公司 Data processing system in network live broadcast

Patent Citations (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105871933A (en) * 2016-06-22 2016-08-17 腾讯科技（深圳）有限公司 Virtual asset allocation system, virtual asset allocation method and virtual asset allocation device
CN107197331A (en) * 2017-05-03 2017-09-22 北京奇艺世纪科技有限公司 A kind of method and device of real-time monitoring live content
CN108932451A (en) * 2017-05-22 2018-12-04 北京金山云网络技术有限公司 Audio-video frequency content analysis method and device
CN111182314A (en) * 2018-11-12 2020-05-19 阿里巴巴集团控股有限公司 Live stream processing method and device and data processing method
CN109766472A (en) * 2018-12-28 2019-05-17 广州华多网络科技有限公司 Signal auditing method, device, electronic equipment and storage medium
CN109831698A (en) * 2018-12-28 2019-05-31 广州华多网络科技有限公司 Signal auditing method, device, electronic equipment and computer-readable storage medium
CN109831459A (en) * 2019-03-22 2019-05-31 百度在线网络技术（北京）有限公司 Method, apparatus, storage medium and the terminal device of secure access
CN110990631A (en) * 2019-12-16 2020-04-10 腾讯科技（深圳）有限公司 Video screening method and device, electronic equipment and storage medium
CN111432227A (en) * 2020-03-27 2020-07-17 广州酷狗计算机科技有限公司 Virtual resource transfer risk determination method, device, server and storage medium
CN111753520A (en) * 2020-06-02 2020-10-09 五八有限公司 Risk prediction method and device, electronic equipment and storage medium
CN111724069A (en) * 2020-06-22 2020-09-29 百度在线网络技术（北京）有限公司 Method, apparatus, device and storage medium for processing data
CN111770353A (en) * 2020-06-24 2020-10-13 北京字节跳动网络技术有限公司 Live broadcast monitoring method and device, electronic equipment and storage medium
CN111770352A (en) * 2020-06-24 2020-10-13 北京字节跳动网络技术有限公司 Security detection method and device, electronic equipment and storage medium
CN111711831A (en) * 2020-06-28 2020-09-25 腾讯科技（深圳）有限公司 Data processing method and device based on interactive behavior and storage medium
CN112199640A (en) * 2020-09-30 2021-01-08 广州市百果园网络科技有限公司 Abnormal user auditing method and device, electronic equipment and storage medium
CN112312152A (en) * 2020-10-27 2021-02-02 浙江集享电子商务有限公司 Data processing system in network live broadcast

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114205676A (en) * 2021-12-08 2022-03-18 广州方硅信息技术有限公司 Live broadcast monitoring method, device, medium and computer equipment
CN114205676B (en) * 2021-12-08 2024-05-28 广州方硅信息技术有限公司 Live broadcast monitoring method, live broadcast monitoring device, live broadcast monitoring medium and computer equipment
CN114679600A (en) * 2022-03-24 2022-06-28 上海哔哩哔哩科技有限公司 Data processing method and device
CN114913172A (en) * 2022-07-13 2022-08-16 广东电网有限责任公司佛山供电局 Method, system, equipment and medium for identifying manufacturing risk of cable middle head
CN116072123A (en) * 2023-03-06 2023-05-05 南昌航天广信科技有限责任公司 Broadcast information playing method and device, readable storage medium and electronic equipment

Similar Documents

Publication Publication Date Title
CN113766256A (en) Live broadcast wind control method and device
US9679074B2 (en) Social genome
US9553918B1 (en) Stateful and stateless cookie operations servers
CN106919579B (en) Information processing method, device and equipment
CN110210883B (en) Group control account identification method, device, server and storage medium
US11756075B2 (en) Systems, methods, and media for detecting suspicious activity
CN107528818B (en) Data processing method and device for media file
CN107077498B (en) Representing entity relationships in online advertisements
US20210390181A1 (en) Generating Simulated Spear Phishing Messages and Customized Cybersecurity Training Modules Using Machine Learning
US10956522B1 (en) Regular expression generation and screening of textual items
CN106533893B (en) Message processing method and system
TWI807172B (en) Active risk control method and system based on intelligent interaction
CN107634947A (en) Limitation malice logs in or the method and apparatus of registration
CN107146012A (en) Risk case processing method and system
US9942255B1 (en) Method and system for detecting abusive behavior in hosted services
CN110401636A (en) A kind of big data air control method and apparatus for supervising abnormal access
Chester et al. The digital commercialisation of US politics—2020 and beyond
CN114186626A (en) Abnormity detection method and device, electronic equipment and computer readable medium
CN102833111A (en) Visual hyper text transfer protocol (HTTP) data supervising method and device
CN108804501B (en) Method and device for detecting effective information
CN113326375A (en) Public opinion processing method, device, electronic equipment and storage medium
CN113746790A (en) Abnormal flow management method, electronic device and storage medium
CN112307464A (en) Fraud identification method and device and electronic equipment
CN109308298B (en) Engineering project integrated management system
CN110336826B (en) Method, device, equipment and storage medium for acquiring interface parameter types

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