Article

IoT decision making improved with impact-sourced human experts

    Author:
  • Deepak Puri, Chief Solutions Architect, SkilledAnalysts.com
  • June 20, 2016
IoT decision making improved with impact

Drowning in data is a real hazard with the Internet of Things (IoT). How should decisions be made with this flood of sensor data? A hybrid approach combining human intelligence and computing power works well.

People are good at making decisions that require nuance and judgement, such as identifying hate speech in online postings. Computerized analytics is better at quickly processing large volumes of data. How do you combine the human thought-making process with the scalability of computing power? In machine learning, this is called supervised learning, where a computer program is taught to “mimic” the thought making-process of a human expert.

Consider an IoT service where sensors on a dog's collar send the pet's vital signs to the cloud for health recommendations. Human experts make recommendations based on the sensor data and factors such as the pet’s breed, weight and age. Computerized decision rules to automate the process are then developed following the same thought process the experts used.

Human Guidance for IoT Data Analysis with Impact Sourced Workers http://www.SkilledAnalysts.com
Factors considered in making a recommendation.

Crowdsourcing

Finding workers to handle such tasks is the next challenge. The number of workers needed for data analysis is hard to estimate, as it depends on the number of customers. Hire too few workers, and customer service suffers. Hire too many, and you have idle workers. Crowdsourcing offers a global pool of workers through an online platform. Workers can be hired to handle online tasks when needed. This provides flexibility and cost savings when responding to fluctuating demands.

Amazon’s Mechanical Turk and Upwork are two crowdsourcing platforms that connect contractor workers with employers that have tasks to be done. A common concern with crowdsourcing, though, is that workers will readily skip from one job to another that pays more. This makes it risky to invest in worker training and harder to ensure deadlines are met.

impact sourcing process for IoT Data Analytics

Impact sourcing

Impact sourcing is socially responsible crowdsourcing that addresses these concerns through dedicated workers. Impact sourcing firms train and employ workers in underemployed communities, such as U.S. veterans, military spouses and African students. These workers are employed by the impact sourcing firms; they aren't freelancers. This motivates them to provide higher quality work and to adhere to deadlines. In our dog sensor example, trained workers analyze sensor data and factor into account the pet’s breed, age and activity level to make diet recommendations.

Other tasks that are frequently impact-sourced include:

  • Digitization—converting materials and tagging content for ebooks
  • Data services—data entry and verifying customer addresses
  • Research services—data collection, market research and customer surveys
  • Image processing—retouching photos, cropping and image enhancement
  • Back office services—customer support and accounting
  • Machine learning—structuring data to help train machine algorithms

Microsoft, eBay and TripAdvisor already use impact-sourced services.

“Microsoft actively supports impact sourcing as an innovative approach to provide both business and societal value. It strongly aligns with our mission to help businesses and people reach their full potential,” explains Tim Hopper, Responsible Sourcing Manager at Microsoft.

Impact sourcing firms

Digital Divide Data (DDD) has a staff of over 1,200 across North America, Asia and Africa, serving hundreds of clients. It’s U.S.-based Liberty Source division offers jobs and opportunities to military spouses and veterans. Since 2001, it has helped lift hundreds of families out of poverty.

iMerit delivers on-­demand data services while creating digital jobs in underserved communities in India. iMerit employs over 600 youth and women to provide machine learning, data and content services to customers such as eBay.

Cloud Factory provides an API-driven service to enable its workers to be integrated online into a client’s operations. Their workforce consists of both employees and local workers who get the flexibility to also attend school and serve in their communities.

Consider impact sourcing for help in handling routine tasks in your next IoT project. The workers can ensure a smoother service launch and also assist your programmers in developing the “business rules” to automate tasks. You'll deliver a better IoT solution and support an underemployed community with work at the same time.


This post was republished from Network World as part of NetHope's effort to facilitate collaborative learning and community knowledge-sharing. Please click here to read the article in its original form. We are always looking for relevant and thought-provoking ICT-related posts to republish. We value your suggestions; if you'd like to recommend a post, please write us at solutions.center@nethope.org.

Our website uses Cookies to enhance your experience and to measure, understand, and improve our website. By continuing to use this website you agree to the use of Cookies. If you do not agree, you can decide to disable Cookies in your web browser. Learn More About Cookies
Got it!