Product Overview

1.What is Text Review?

The LiveData text review system is based on massive sample data, combined with deep learning and natural language processing technology, to detect and identify illegal content such as pornography, politics, violence, terrorism, hate speech, and advertisements contained in texts.

2.Product Features

The text review system supports the detection and identification of the following types of illegal content:

Violation Type Describe
Politics Accurately identify domestic and foreign political, anti-party and anti-government, sensitive persons, events and other illegal texts.
pornography Accurately identify explicit expressions about sexual behaviors, sexual organs, and illegal texts about pornographic insinuations, transactions, and dissemination.
Abuse Accurately identify illegal texts such as abuse, attacks on others, and uncivilized language.
Violent Accurately identify illegal texts containing relevant content such as ISIS and other terrorist organizations, violent terrorist incidents, and terrorist speeches.
Prohibited Accurately identify illegal texts that contain prohibited content such as propaganda of cults, feudal superstitions, guns and ammunition, drugs, etc.
Hate Speech Accurately identify illegal texts containing racial discrimination, sexual orientation discrimination, etc. that are easily controversial.
Advertising Accurately identify junk content in games such as malicious screen-swiping ads, micro-business ads, and game-pulling ads.
Protection of Minors Accurately identify content that directly or indirectly expresses that the speaker is under the age of 18 and contains the meaning of attracting child pornography.
Personal Information Protection Accurately identify content that discloses personal information such as ID card, phone number, offline address, ip, email address, etc. in the text.
Private Business Accurately identify the transfer information, online payment tools, etc. in the text that involve transactions between individuals.
Sensitive hotspot Accurately identify social hotspots and events in the text that need to be blocked.
Irregular Emoticons Accurately identify illegal texts that express sensitive meanings through expressions.
Nickname Accurately identify content suspected of impersonating official staff, system announcements, etc,in user nicknames.
Custom content Support users to define their own keyword library, and perform targeted filtering of sudden harmful content in emergency response scenarios.

3.Service use process

  1. Click “Register now” to register LiveData( https://www.ilivedata.com/

  2. Create a company (fill in the company name, contact person and other information, wait for review)

  3. Create a project in the console under “Text Review” service (fill in project name, project description, etc.)

  4. Interface debugging (get the project ID and key, and debug the interface).

4.Application Scenes

Including but not limited to chat, comments, barrage, signature/nickname, AI-generated dialogue and other scenarios.

We support real-time detection and identification of chat messages such as P2P, rooms, groups, alliances, reviews of stores and products, barrage content in videos and live broadcasts, and personal signature/nickname content in games, social platforms, etc.; we support timely and effective interception of illegal and illegal information, protecting users from malicious harassment and infringement, creating a good network environment, and avoiding platform supervision and operational risks.

AIGC scenario: We support splicing and detection of text correlation semantics of AICG large models in various scenes at home and abroad. After turning on splicing detection, the speech content with the same sessionID and userID will be sent for correlation streaming detection within the specified interval.