How Sigma Works
Our senior team has over 30 years experience in Data Annotation. It was in the early 90’s when they developed the first specific user-friendly tools for massive and collaborative data annotation in the world, reducing the annotation time by 85% while keeping accuracies over 98%.
Since the creation of Sigma in 2008, the team has continued to work in improving the data annotation processes and tools to achieve even higher efficiency, data quality and security. In particular, the development of Machine Learning (ML) and Digital Signal Processing (DSP) assisted tools together with the unique approach to tools development has helped significantly improve all the Key Performance Indicators (KPIs).
In Sigma we recognize that each client is unique and has its own specific objectives, strategies and procedures. We also recognize that each data collection or data annotation project has its own particularities that need to be analyzed and taken into account.
It is for these reasons that Sigma’s approach is very flexible. Whether we use the client tools and procedures or we use our own, whether we provide the services remotely, on premise or in our highly secure facilities, we’ll go to the end of earth to provide the best service possible. This includes the adaptation of our human and technology resources as well as our work procedures to the client and the project.
Nevertheless, our standard approach to data collection and annotation is the following:
The data collection/annotation project is assigned to the project manager whose experience matches the project type. The Project Advisory Committee (PAC) is created. The PAC is composed of experts in the corresponding project area. For example, speech technology, computer vision or data engineers, computational linguists, tools development engineers, quality assessment experts, etc.
The entire system is tested and refined.
The Project Advisory Committee prepares a report with results and suggestions.
The report and the sample data is delivered to the client.
If client has updated the guidelines, the tools are updated accordingly.
Annotators start producing data at scale. The iterative process to improve performance of the AI-Assisted tools starts.
The Quality Assessment methodology is not applied linearly. Annotators who have just started to work in the project are more closely monitored a supervised until they perfectly interpret and apply the guidelines and their level of quality reaches the project objectives.
Annotation Service Options
|Workforce||NDA, Security & Privacy||Data Access Controls||Additional Security Measures||Additional Information|
|Remote||Sigma Global Workforce||Yes||SSL, VPN||N/A||Annotators work online from home|
|On Client’s||Sigma’s Local Workforce||Yes||N/A||N/A||Annotators work in client’s office|
|Sigma´s Office||Sigma’s Local Workforce||Yes||SSL, VPN, 2FA||24×7 video cameras||Annotators work in Sigma’s office|
|Sigma’s highly Secure Facilities||Sigma’s Local Workforce||Yes||SSL, VPN, 2FA||24×7 video cameras, security restrictions (*)||Annotators work in Sigma’s Highly Secure Facilities|
|Customized Highly Secure Facilities||Sigma’s Local Workforce||Yes||SSL, VPN, 2FA||24×7 video cameras, security restrictions (*)||Sigma is experienced in designing, implementing and operating Secure Facilities|
(*) For more information on security measures and restrictions, please, contact us.