Magazine Store

10 Best Artificial Intelligence Companies to Watch 2020

Fuel your AI applications and initiatives with high quality training data: DefinedCrowd


As AI technologies mature, it’s becoming more and clearer every day that data quality is the fuel that drives models to sustained success.However, most organizations’ frameworks for selecting data providers still don’t take quality into account.At DefinedCrowd, privacy and security are of utmost importance. You can rely on its comprehensive and robust security practices, whether it is internal security, data protection and encryption, or security compliance.Whether you have data or not, DefinedCrowd’s SaaS platform allows you to create a project, set data parameters and monitor progress.They use human intelligence combined with machine learning algorithms to validate the quality of your data, using several verification processes and quality threshold guarantees.At the end of this process, you receive enriched and structured data ready to start training your models right away.

The API provides a programmable interface that directly integrates your data processes with our platform. You can create a project based on Workflow Templates with or without previous data, and track performance during execution. Upon completion, you’ll be able to quickly retrieve high-quality training data enriched by our platform.DefinedCrowd offers an intelligent data infrastructure that provides high-quality training data that enables customers with AI roadmaps reach market at better quality and speed. DefinedCrowd collects, structures, and enriches high-quality training data for AI initiatives using efficient data workflows that combine human intelligence, automatic tools and machine learning capabilities. Customers may submit their data for evaluation or request that DefinedCrowd collect training data on their behalf.

This REST API provides a programmable interface for directly integrating your data processes with DefinedCrowd’s platform. From the REST API, you can create Projects based on Workflow Templates, upload your data, start Projects, and manage them during execution. Upon Project completion, you’ll be able to quickly retrieve high-quality training data enriched by the DefinedCrowd platform.Here are some essential concepts to help you understand and use our REST API to its fullest potential:

  • A Contributor is someone who participates in the project execution. This is our human intelligence component.
  • One or more Contributors work on Input Units. When you submit data for evaluation, each individual item within that submitted data is an Input Unit.
  • A Workflow Template is a pre-configured set of steps designed for executing specific types of tasks. A Project combines a Workflow Template with Input Units.
  • A Project’s configuration determines the rules of execution for each input unit (e.g. how many times each input is executed by each Contributor). Depending on the Workflow Template, different steps and configuration options are available.

Workflow Templates can be data collection specific. Here, DefinedCrowd uses data-based “prompts” that you provide as Input Units and are then displayed in our data collection tools. DefinedCrowd also provides Workflow Templates for structuring and enriching customer-provided data. That data, which you provide as Input Units, can be text, audio or images.

DefinedCrowd’s REST API is based on four primary actions:

  1. Create a Project – A Project consists of a specific Workflow Template, a configuration that instructs DefinedCrowd how to execute the Project and the actual data to operate on.
  2. Upload Input Units – This is a unit of work that can be performed by one or more Contributors. Each uploaded unit within the data that you submit will be validated and uploaded into the platform. Each unit can then be executed multiple times by different Contributors in different steps, depending on the configuration defined in the project.
  3. Start Project – Starts the Project. When a project is started, you are charged for the project and execution begins.
  4. Get Results – When project execution is complete, this action allows you to obtain the results.

Any application utilizing DefinedCrowd’s REST APIs must first authenticate with its service. Their service utilizes the OAuth v2 protocol for authenticating. This requires both an “Access Key ID” and an “Access Key Secret” that can be secured through DefinedCrowd’s Enterprise Portal. Associated with an Access Key pair is a set of permissions which are scoped to different functionalities. Select the appropriate permissions in the Enterprise Portal that support your scenario when creating the Access Key pair, as these permissions are bound to the Access Key pair. The Access Key pair must be secured appropriately and cannot be shared.In addition to successfully authenticating with DefinedCrowd’s service, an application must have the required Permissions to invoke an API. In the documentation below, for each REST API endpoint the permission required is described. Associated with each “Access Key ID” and “Access Key Secret” that is secured through DefinedCrowd’s Enterprise Portal is a set of Access Key permissions that are an aggregate of the permissions required for each REST API. Based on your scenario, review the REST API’s below, identify the permissions required and create the appropriate API Key permissions.

By design, the DefinedCrowd platform scales large workloads across its data processes. Their platform is largely asynchronous. For high-quality results, it may take several days to complete a project. As a result, we provide an event-based model using webhooks. We strongly advise registering a webhook that responds to events / callbacks and developing your solution to take actions based on the events received.A URL reference to a service created and managed by the customer is registered on Create project, allowing those customers to receive Project State change events.

Daniela Braga Founder & CEO of DefinedCrowd

Daniela is the creator and driving force behind DefinedCrowd, with over eighteen years’ experience in speech technology in both academia and industry. Alongside her role as our CEO, she is also a regular guest lecturer at the University of Washington and author of more than 90 scientific papers and several patents. Previously at Microsoft, she was involved in the creation of Cortana. She then formed the Data Science team for Voicebox Technologies where she introduced crowdsourcing for big data solutions. She has a deep passion for people and enjoys keeping active through hiking, skiing and learning new things, from the latest technology to how to dance or sail.

“Our mission is to accelerate the evolution of AI initiatives with a quality-focused data platform that combines the best of machine learning and human intelligence. Meet DefinedCrowd, your partner in data.”