The Ground Truth Grind: Strategies for Measuring the Quality of Labeled Natural Language Data

The Ground Truth Grind: Strategies for Measuring the Quality of Labeled Natural Language Data

Wednesday, June 17, 2020 7:30 PM to 8:00 PM · 30 min. (Africa/Abidjan)
VG5
Talk

Information

Tired of spending precious time and resources on the ground truth grind? Join iMerit Solutions Architect, Teresa O'Neill, who discusses:

  • Three cornerstones for building trust in the quality of your labeled natural language datasets
  • Strategies for determining what “high quality” means in the context of your project
  • Quality control process and metrics design that is right for your use case"
iMeritOur Teresa O'Neill, PHD is presenting this afternoon in VG5: "The Ground Truth Grind: Strategies for Measuring the Quality of Labeled Natural Language Data" Most language data goes unused: it is time-consuming to annotate, and current NLP and AI models do not perform well in new contexts. Skilled annotation extracts hidden insights from audio and text to unlock the power of natural language data in new business contexts. Here are some of the use cases we’re working on: Audio Validation & Transcription Sentiment & Intent Analysis Named Entity Recognition & Linking Topic Classification Grammatical & Discourse Analysis iMerit provides high-volume data labeling and enrichment services using our in-house workforce of 2,500+ full time data technicians located domestically and abroad. Our internal SMEs, who have their PhDs in Linguistics, scale and oversee all of our text and audio projects. Whether you’re just starting out or are well underway, we would love to learn about your use case and explore the ways we can potentially support your team. Be sure to check out our very own Teresa O'Neill, PHD as she presents in VG5 this afternoon! 

Registered attendees

Akanksha Bhasin
Akanksha Bhasin
Conversational Designer & Community AdvocateStudent
Denis Ardabatsky
Denis Ardabatsky
CEONeoSound

Log in