With the rapid development of artificial intelligence (AI) industry, the development of AI data becomes more and more important.
Insufficient security protection of traditional AI data
Data mining is the prerequisite for obtaining AI training samples. With the development of AI data service industry, data annotation becomes a hot industry. Many AI startups use the current mainstream model of data annotation industry – “outsourcing”. “Outsourcing” can be further divided into two models: “crowdsourcing” and “factory”.
However, due to the high labor cost, in many cases, the “factory” has to outsource some of its business to smaller teams. Therefore, in addition to these relatively standardized institutions, there exists many “small workshops”. They both have disadvantages: “small workshops” do not have annotation tools, while “factory” operates in a more streamlined way but lacking of practical operation model. Considering the large number of workers involved in annotation work, and the unregulated process of data annotation and delivery, data security becomes a major hidden risk.
Introduction of Security related regulations and polices
In Feb. 2014, Leading Group of Cyberspace Affairs of CPC Central Committee, headed by Xi Jinping, was established. In 2016, the State Council issued <"Thirteenth Five Year" National Information Technology Plan> and pointed out that "the state plans to implement big data security protection project, to enhance the security protection of data assets during transmission, storage, usage and publication. In March 2018, the Leading Group of Cyberspace Affairs was restructured, and Central Cyberspace Affairs Commission was established. The central government's emphasis on cybersecurity is the major driving force.
Efforts of Datatang on AI data security
AI data security threats and demands, as well as the legal protection have jointly promoted the development of AI data security technology. In the 15th IEEE International Conference on Mobile Ad-hoc and Sensor Systems (IEEE MASS 2018), Datatang AI Lab and Beijing Institute of Graphic Communication published the paper "A Framework to Data Delivery Security for Big Data Annotation Delivery System", focusing on the research and application of AI data transmission security. This paper has received extensive attention from the industrial insiders.
Datatang, as a professional AI data service provider, offers services covering the entire life cycle of AI data, including data collection, data annotation and data exchange, as well as data exchange with advanced independent development tools for various types of data. Data transfer is the last critical step in sending data to customers. One of the principles is to provide security management to avoid theft, exposure and duplication. Most of the security issues at this stage are due to incorrect, inappropriate or deliberate human operations. Therefore, an automated delivery system is requested to reduce human intervention.
“Data Online Delivery and Acceptance System” is the best application case of Datatang based on its data security technology. The system is a value-added service system tailor-made for customers from their point of view. Traditionally, data is delivered via email, network disk or hard disk. It lacks security and is troublesome for check and acceptance. Furthermore, it is very costly to follow up on the delivery progress. The online delivery and acceptance system developed by Datatang allows customers to log in to follow project delivery progress, perform real-time acceptance, provide online feedback, and download data directly. It solves the problems of traditional data delivery, enables convenient communication, timely acceptance, flexible and safe delivery.
Further development with original aspiration
Founded in 2011, Datatang (Beijing) Technology Co., Ltd. is a technology company specializing in AI data service, providing data acquisition and data product services to AI companies all over the world. Headquartered in Beijing, the company has eight wholly-owned and controlled subsidiaries in China and a US subsidiary in Silicon Valley. Currently, it has several professional data processing centers in Nanjing, Baoding, Hefei and Guiyang. The company’s data collection business has extended to more than 30 countries around the world, with partners in more than 10 countries.
In recent years, with the rapid development of AI technology, the increase of AI data set has exploded, and higher requirements on data transmission capability of the data transmission system have been put forward. It has been the mainstream development trend to take into consideration of data transmission capability and security at the same time. Datatang will continue to increase R&D investment, vigorously promote technological innovation, and advocate standards and norms on AI data security management, while maximizing the value of data to promote innovation on AI technology, applications and production, therefore, contributing to the rapid development of AI industry.