SHANGHAI, March 18, 2022 /PRNewswire/ — IDC recently analyzed the status quo of enterprises in the development and construction of AI applications, and summarized the characteristics of data services required for current AI model training in rising to data service challenges and needs. To tackle the development trend of AI data services, Appen, a world-leading AI training data service provider, now provides data solutions featuring unique advantages for the entire AI lifecycle in the Chinese market.

1. AI is being employed and implemented across a wide range of industries with the rapid development of AI market in China.
2. Enterprises find their data service needs are changing dramatically as they need to handle privacy concerns, the lack of data, and other challenges in adopting AI.
3. To accelerate the process, they are attaching increasing importance to AI data, as well as the quality, efficiency, and safety of AI data services rendered.
4. As a world-leading AI training data service provider boasting unique advantages in the Chinese market, Appen furnishes enterprises with full-stack acquisition and annotation services and solutions for images, texts, voice, audio, and video in the AI lifecycle. It provides OTS dataset products and intelligent data annotation On-Promise/SaaS/Hybrid platforms to be more responsive to the various needs of customers, manifesting outstanding advantages in the fields of foreign languages and autonomous driving data services.

AI Market in China

In recent years, the AI data service market has been fueled by the increasing need for rich and high-quality data sources brought by the rapid development of AI market in China. As shown in the figure below, IDC predicts that in 2021, the overall size of AI market in China will reach 8.22 billion US dollars (about one-fifth of the counterpart in the United States), and it’s expected to reach 16.3 billion US dollars by 2025. IDC forecasts that the five-year compound annual growth rate (CAGR) of the AI market in China will surpass that of the United States and the rest of the world in the post-pandemic era.

Enterprises are shifting their exploration of artificial intelligence technologies from single applications to multiple business scenarios and evolving from informatization to intelligentization, a more advanced stage, as the digital transformation enters "deep-water areas". Enterprises begin to employ AI-generated audio content, knowledge graphs, multi-modal interactions, and other technologies in their production environments. The application scenarios of such technologies have been expanded from simple identity authentication and intelligent customer services to intelligent process automation, meeting assistants, intelligent writing, virtual assistants for employees, AI digital humans, and so on. In terms of business effects, the business value of AI applications is becoming clearly visible and even assessable.

With a predominant value-added effect, AI has promoted the intelligent transformation of all walks of life, yet the implementation process is fraught with challenges. Problems occur in strategies, talents and teams, data, skills, processes, and many other areas. In a survey conducted in 2021, IDC found that the top three challenges faced by enterprises worldwide when adopting AI were high upfront costs, the lack of machine learning operations (MLOps) expertise, and the lack of data science expertise. Privacy concerns and the lack of training and testing data were also too serious to be ignored.

The widespread introduction of deep neural networks into industry applications has raised massive data needs, yet business effects cannot be significantly improved by continuing to optimize the model codes after such models are relatively mature. Therefore, optimizing the training data has become an important means for greatly improving the accuracy of AI models.

In this context, enterprises are devoting more resources to the acquisition of high-quality training data, not just model training, in their AI R&D. IDC found that, of the global enterprises that have been interviewed, 85% believe they spent more than half of their AI development investment to data preparation.

Appen China MatrixGo Data Annotation Platform

The MatrixGo high-precision data annotation platform is a platform product that Appen independently developed after entering the Chinese market. It includes Appen’s overseas practical experience, yet also fits the characteristics of the local market. Integrating rich and efficient annotation tools, it has powerful workflow scheduling capabilities and can support various annotation projects in rich scenarios with massive data by incorporating resources from its teams, supplier teams, and millions of crowdsourcing personnel. The platform can be delivered through private deployment, software as a service (SaaS), or hybrid cloud deployment, where SaaS can be integrated with privately deployed file service to further ensure compliance of data flow and strengthen data security management.

As a professional segmented solution-level product platform, it can provide functions for all core scenarios in the project execution process, including data requirement proposal and strategy formulation, data collection, high-precision annotation/classification, progress tracking and quality control, and full data delivery. It effectively integrates AI and model capabilities, enables large-scale man-machine coordination, and supports massive data acquisition, multilingual and multi-scene speech transcription and translation, complex content relevance evaluation, computer vision target recognition and tracking, and 3D object tracking in point clouds, semantic segmentation, and other functions to provide customers with high-quality AI training data in e-commerce, online social networking, content services, smart healthcare, smart hardware, autonomous driving, and other fields.

Appen’s MatrixGo platform has delivered the acquisition and annotation services for various AI applications, supporting and serving more than 130 customers and 700 projects in China in 2021.

Best Practices

1. It’s a Chinese listed company specialized in intelligent voice and AI. As a typical representative of AI technology providers, it enjoys substantial shares in AI voice and semantics markets in China. It has a large demand size for data services, and about half of its expenditure is spent on external services in the form of integrated acquisition and annotation. The other half of demand is processed internally, for example, through its internal crowdsourcing platform.
   Why they chose Appen: 1) The efficiency of communications on the standards of data acquisition and annotation services is high; 2) Appen responds quickly during the project cooperation. 3) The quality evaluation results are good.
2. As a typical representative of Chinese smartphone manufacturers, it reported that the number of global monthly active users of its mobile operating system and content ecosystem was about 400 million in 2021. With the expansion of business scenarios, it found that the needs for data services were growing rapidly. As its internal annotation team could not meet the business needs, it began to cooperate with Appen in 2021. Data annotation services it purchased have covered four domains, namely NLP content understanding, AI assistant voice recognition, computer vision (CV) picture or video taking, and the Internet of things (IoT) device sensors.
   Why they chose Appen:1) The high-quality annotation could meet project progress requirements. 2) Appen’s relatively comprehensive business coverage could meet the requirements of NLP, CV, automatic speech recognition (ASR), and other businesses. 3) Appen has demonstrated a good service attitude. It worked actively to solve new problems after the process for new projects was established and strove to improve project delivery quality.

The demand for data resources in AI development is driving the evolution of the data service market and bringing many opportunities. Nowadays Appen is providing tens of millions of data per month. It serves autonomous driving, technologies and the Internet, healthcare, finance, and other industries in the Chinese market, and features unique advantages for the entire AI lifecycle in the market.