Researchers at the University of California, San Diego wrote a simple script that identifies figurative phrases and replaces them with their literal meaning, as a means of improving the performance of dialogue systems like chatbots.
The researchers found performance declines 10% to 20% when such systems encounter dialogue that incorporates idioms or similes, but their partial remedy can boost the performance of dialogue systems by up to 15%. They describe their work in "Investigating Robustness of Dialog Models to Popular Figurative Language Constructs," presented at the 2021 Conference on Empirical Methods in Natural Language Processing.
The script allows the systems to check dictionaries that translate figurative speech into literal speech, which the researchers found to be more efficient than retraining the systems to learn the complete content of the dictionaries. "We want to enable more natural conversations between people and dialogue systems," says Harsh Jhamtani, a Ph.D. student at Carnegie Mellon University.
From University of California, San Diego
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found