How to Label Data Quickly in a New Domain

In this episode, we chat with Luke Zettlemoyer about Question Answering as a format for crowdsourcing annotations of various semantic phenomena in text. We start by talking about QA-SRL and QAMR, two datasets that use QA pairs to annotate predicate-argument relations at the sentence level. Luke describes how this annotation scheme makes it possible to obtain annotations from non-experts, and discusses the tradeoffs involved in choosing this scheme. Then we talk about the challenges involved in using QA-based annotations for more complex phenomena like coreference. Finally, we briefly discuss the value of crowd-labeled datasets given the recent developments in pretraining large language models. Luke is an associate professor at the University of Washington and a Research Scientist at Facebook AI Research.