OSU graduate students Nikhita Vedula and Pranav Maneriker received the Best Paper award at the Web Conference 2020 (formerly WWW) with their paper “Open Intent Extraction from Natural Language Interactions,” co-authored by TDAI affiliate Srini Parthasarathy and Nedim Lipka from Adobe.
The conference, which occurs annually, focuses on analyzing the future of The World Wide Web. Participants are eligible to submit and receive awards in categories such as Best Paper, Best Student Paper, Best Poster, the Seoul Test of Time Award and an Honorable mention. This year, the event was intended to take place in Teipei, Taiwan from April 20th to April 24th; however, due to concerns of safety regarding the COVID-19 public health crisis, it was instead held online.
Vedula and Maneriker’s paper focuses on how user intent derived from written or spoken language is essential in promoting natural language understanding as well as automated dialog response. They propose utilizing OPINE (OPen INtent Extraction), a new domain-agnostic approach, to formulate “the problem as a sequence tagging task under an open-world setting.” Through countless experiments, OPINE has been tested and has shown its efficacy in various domain intents despite little to no training examples for each different domain.
Congratulations to this team for such a wonderful achievement.