PNNL

“Corpus-based and Knowledge-based Measures of Text Semantic Similarity”  was coauthored by Courtney Corley, a PNNL data scientist and group leader, and was recently selected for a 2022 Association for the Advancement of Artificial Intelligence (AAAI) Classic Paper Honorable Mention. It was dubbed the most influential paper from the Twenty-First National Conference on Artificial Intelligence and was specifically honored for its “pioneering work to use word semantics to assess text similarity.”

The AAAI Classic Paper award was established in 1999 to honor the author(s) of papers deemed most influential from a given conference year.

Corley’s study helped shed light on how computing systems can better understand human language. At the time of its publication in 2006, the paper offered a state-of-the-art methodology for combining various machine learning techniques to derive bigger picture insights into the words we use and the stories they tell. Ultimately, equipping computers to understand what humans mean when they communicate has far-ranging implications. Some of the applications include developing more effective chatbots and conversational agents, understanding topics like disease spread or public sentiment, creating systems that help us translate languages without losing context, and more.

To this day, Corley’s research remains the most highly cited paper from the year it was published with over 1,600 citations.

Corley’s expertise in artificial intelligence and biosurveillance has allowed him to make foundational contributions to fields, such as social media analytics, computational epidemiology, and natural language processing. Most recently, he has been focused on developing machine learning approaches to solve challenges in limited data regimes, such as few-shot learning. Separately, he has been studying how to create trustworthy artificial intelligence systems that are safe, secure, and robust.

The classic paper award will be presented during the AAAI virtual ceremony at AAAI-22.