Content security has become a global Internet ecosystem management problems. Internet platform of multimedia content to bring the outbreak of the mass of information, but also pays top academics coerced a large number of bad harmful information.
When the content platform rapid development, will be eyeing production of gray, regarded as "flow honeypot", to use its flow exposure opportunity, implants involving pornography, gambling and other undesirable information, and a lot of junk advertising. According to incomplete statistics, in 2019, there are thousands of apps and content security services because of problems encountered while shutting down the shelf and other reform, business suffered huge losses.
December 19, 2019, NetEase easy to shield successfully held in Beijing 2019 conference content security intelligence system, forward thinking and easy to build a shield to share content security system, and officially released audit management system for smart era, there are hundreds of companies content-related risk control personnel to participate in the event, to discuss the contents of the safety system construction and practice.
According to the survey, harmful spam accounted for about 3% of the amount of Internet UGC content. Most of which are all kinds of illegal advertising category, followed by Jurisprudence vulgar class information, gambling and drug-related information. With the integration of multimedia content, 5G and other information infrastructure, Internet content has also ushered in a large outbreak, content security even more severe pressure.
Netease Zhu Hao Qi Yi shield CTO
Netease Zhu Hao Qi Yi shield CTO believes that only industry efforts together to build a smart driver, unified linkage, personalized custom build all the content system of governance, in order to fully and efficiently solve the problem of content security.
-
Intelligent drive: the next-generation content security system, intelligent technology plays a role in driving the engine. Currently more than 90% of the mass real-time multimedia content can be filtered intelligence technology accurately identified. Natural language, image recognition, voiceprint identification so thoroughly ground in a mature content security.
-
Unified linkage: Content security needs completely resolved from the source, you need to account security linkage, all-Link linkage recognition analysis. The use of anti-cheat technology verification code, registration protection, login protection, content security cloud linkage capabilities and local audit management system, it can more effectively reduce the harmful content.
-
Personalized custom: different content regulations in all areas of live short video, literature, social media, etc., need the flexibility to customize their own content security policies based on product form.
-
Build public governance: harmful content is a social problem. Companies need to effectively integrate a variety of government forces legal, technical services, peer alliances, joint control of the user, etc., in order to effectively combat the gray production, build positive Internet ecosystem Delicate gas.
此次，网易易盾还发布了面向智能时代设计的审核管理系统。
在未来很长时间内，内容安全依然是人机协同模式，各大平台需要保有一定规模的人工审核部门。但易盾在调研时，发现由于审核管理长久一直缺乏相应通用规范，操作后台长期得不到研发支持，大多数审核管理系统存在操作繁琐、设计陈旧、功能落后，甚至有些操作需要在多个后台跳转处理的问题。
相比较爆发式的内容增长和日趋严格的监管要求，审核管理系统已成为企业内容安全的最大短板，严重影响了审核效率和安全要求。
网易易盾此次发布的审核管理系统是面向智能时代的审核机制而设计，具有功能丰富、智能高效、轻量灵活、交互友好四大优势。
-
功能丰富：支持面向内容的预处理、多重审核、质检溯源全生命周期管理；面向不同业务场景如直播音视频、长文档等垂直场景都有专门的视频模式。
-
智能高效：基于机器学习的聚类审核，同类内容及时预警，审核人员可以快速过滤阻止扩散；高并发支持大规模审核团队同时操作，可智能分组排版。
-
轻量灵活：无技术支持也可以快速接入，支持SaaS和私有化定制，客户可以灵活自定义内容接口、安全策略、外观、业务模块等。
-
交互友好：深入审核团队研究交互习惯，优化界面交互、功能排列、操作流程。建立积分勋章等游戏化设计，有效降低审核疲惫。单手操作等无障碍辅助设计，让残障人士也可以轻松操作。
网易易盾产品总监饶晓艳表示：“审核人员是企业内容安全的重要保障力量，为公司健康发展付出了很多，非常高兴可以为审核人员设计一套友好有温度的管理系统，让他们的工作更加轻松快捷。”
网易易盾脱胎于网易安全部，扎根内容安全领域已超过20年，2016年对外商用以来，已服务数千家企业，其中不乏OPPO/VIVO、B站、一直播等知名客户，每日过滤有害信息超过十亿条，目前已是国内领先的内容安全和业务安全服务商。
