当前位置:网站首页>Best practices cloud development cloudbase content audit capability
Best practices cloud development cloudbase content audit capability
2022-06-23 17:47:00 【Tencent cloud development TCB】
This article introduces the usage scenario of cloud development content audit , And demonstrate how to use the content audit function .
* This article mainly introduces the function application of Tencent cloud development side , For the user guide of wechat cloud development side, see :
https://developers.weixin.qq.com/miniprogram/dev/wxcloud/basis/Consecurity.html
Content review introduction
What is content auditing ?
Content review is based on images 、 Text 、 Audio and video detection technology , Can automatically detect vulgar 、 Breaking the law 、 porn 、 advertisement 、 Graphics and text violations, etc , Upload pictures to users 、 written words 、 Audio and video content audit , To meet the upload requirements , Help customers reduce the risk of business violations .
Common content auditing methods include :
- Image content review
Image content review , Using the depth neural network model to detect the image content , Accurately identify the sensitive illegal elements in the image 、 Illegal contents, etc , Help the business avoid the risk of violation .
- Text content review
Text content review , Use artificial intelligence text detection technology to effectively identify violations 、 advertisement 、 abuse 、 Contraband and irrigation text content , Provide customized text sensitive content audit scheme .
Cloud development content audit introduction
The cloud development content audit is the content hosted in the cloud development ( Data stored in the cloud database 、 Files stored on cloud storage ) Provide fast content audit services . Content approval rules can be set quickly ( Audit scope 、 Audit scenario 、 After violation, it will be automatically blocked ), Asynchronous audit after database document changes , The text content is reviewed and processed in seconds .
Be careful : Currently, only the documents in the cloud database can be audited .
Common scenarios for content auditing
In the current Internet Environment , Content auditing can be said to be everywhere , Take a forum website as an example : As small as a user's username registration content audit 、 Users' comments in the comment area or message area are reviewed , To the content review of the entire post posted by the user .
Where users can freely input content, their content needs to be reviewed , So as to prevent the forum from being surrounded by a lot of bad information , The web server resources are wasted , Seriously affected the image of the website , What's more, the webmaster may bear relevant legal responsibilities .
Use the cloud to develop content audits
Prerequisite
There is already one available Pay by volume Environmental Science .
https://cloud.tencent.com/document/product/876/41391
Operation steps
Click on a metered environment , Get into Environmental Science > Content review > Security configuration page , On the security configuration page, you can see that there are two main parts : Audit rules and audit details .
step 1: Create an audit rule
Before content review , You need to create an approval rule first , That is to say, what rules should be formulated to review the contents .
1、 Click New , Start creating audit rules .
2、 Select which collection in the cloud database to audit , For example, users' comments need to be reviewed , Then select a collection to store user comments , as follows :
explain : The sets listed in the drop-down box are all the sets in the database , If there is no set in the current drop-down box , Then you need to create a collection in the left navigation bar database .
3、 After selecting the set , You also need to fill in which field under the set to approve , For example, users' comments need to be reviewed , Then fill in the comment content field in the comment set , as follows :
4、 Draw up a block copy , When there are violations , Information used to prompt the user , The default is : This content has violated the rules and will not be displayed for the time being . as follows :
5、 Turn on the automatic shielding switch for illegal contents . After opening , If the system detects a content violation , The content will be automatically shielded , And 4 Rewrite the user-defined shielding document proposed in . If you don't turn it on , Only the content will be reviewed and scored , Do not override mask automatically . It is generally recommended to turn on , as follows :
6、 After opening , You need to select an audit scenario , That is, what types of violations are reviewed , Such as abuse 、 advertisement 、 It is illegal to 、 Pornography, etc . It is generally recommended to select all , as follows :
explain : The audit score after the violation type refers to : When the system reviews the content , The content to be audited will be scored , The higher the score, the higher the content violation degree , When the score printed is greater than the audit score filled in , The system will automatically screen the content . The audit score can be understood as the evaluation of the current violation type “ Tolerance ”, The lower the score, the lower the tolerance , For example, zero tolerance for illegal information , In line with “ Would rather kill a thousand by mistake , Not one ” Principles , You can fill in the scores of illegal types a little lower ( The minimum 60 branch ), In this way, once the system finds that the content is suspected to be illegal information , Shield it immediately .
7、 After filling in the approval scenario , Finally, click the tick clause .
single click determine . You can create a new approval rule .
step 2: Test audit rules
After the approval rule is created , You can test whether the approval rule created under is effective .
1、 Click the database page... In the left navigation bar , Enter the set selected above , Click add document , Add a normal content first :
Here's the picture , The content without violation will be inserted into the collection normally .
2、 Add the document again , And try to deliberately enter some illegal content :
Here's the picture , The inserted document field contains illegal content , It has been rewritten by auto mask .
The above description indicates that the created approval rule has taken effect .
step 3: View audit details
1、 Click... In the left navigation bar Content review After page , Then click... At the top Audit details , You can see the audit content and audit results of the audit platform , as follows :
2、 When testing audit rules , The illegal content intentionally entered for the second time has been successfully monitored by the audit system , And accurately identify the types of violations and the audit scores of the contents , as follows :
3、 In some specific demand scenarios , If it is found that a certain item belongs to the normal content in a specific scenario , However, the audited system mistakenly determines that the content is illegal :
This violation record can be Roll back .
After rollback, the processing status of this item in the audit record also changes to Rolled back , And the record can be displayed normally in the database .
* This article mainly introduces the function application of Tencent cloud development side , For instructions on the wechat cloud development side, please refer to :
https://developers.weixin.qq.com/miniprogram/dev/wxcloud/basis/Consecurity.html
边栏推荐
- Jetpack compose and material you FAQs
- C. Product 1 Modulo N-Codeforces Round #716 (Div. 2)
- What does the timestamp 90K mean?
- Three functional forms of intelligent switch
- [go] calling Alipay to scan code for payment in a sandbox environment
- FPN characteristic pyramid network
- 如何通过线上股票开户?在线开户安全么?
- Hapoxy-集群服务搭建
- Digital intelligent supply chain collaboration solution for new energy industry
- Date to localdatetime
猜你喜欢

【网络通信 -- WebRTC】WebRTC 源码分析 -- 接收端带宽估计

Self supervised learning (SSL)

DataNode进入Stale状态问题排查
NLP paper reading | improving semantic representation of intention recognition: isotropic regularization method in supervised pre training

混沌工程在云原生中间件稳定性治理中的实践分享

时间戳90K是什么意思?

MySQL事务及其特性与锁机制

Interface ownership dispute
![[go]沙盒环境下调用支付宝扫码支付](/img/d4/c6d72a697bc08f69f11121a15109b3.png)
[go]沙盒环境下调用支付宝扫码支付
![[network communication -- webrtc] analysis of webrtc source code -- supplement of pacingcontroller related knowledge points](/img/18/dda0c7fa33ee6cdf2bbb2f85099940.png)
[network communication -- webrtc] analysis of webrtc source code -- supplement of pacingcontroller related knowledge points
随机推荐
Goframe framework: fast implementation of service end flow limiting Middleware
单火线设计系列文章10:拓展应用-单火开关实现双控
B. Integers Shop-Hello 2022
[go]沙盒环境下调用支付宝扫码支付
AMQP协议
浅谈5类过零检测电路
How to open an account through online stock? Is online account opening safe?
qYKVEtqdDg
Tencent three sides: how to duplicate 4billion QQ numbers?
qYKVEtqdDg
Analysis of object class structure in Nanny level teaching (common class) [source code attached]
Postgresql_ Optimize SQL based on execution plan
Hapoxy-集群服务搭建
解答02:Smith圆为什么能“上感下容 左串右并”?
[go] calling Alipay to scan code for payment in a sandbox environment
I successfully joined the company with 27K ByteDance. This interview notes on software testing has benefited me for life
MySQL - reasons for using repeatable read
股票网上开户及开户流程怎样?在线开户安全么?
What does the timestamp 90K mean?
[network communication -- webrtc] source code analysis of webrtc -- bandwidth estimation at the receiving end