Unlikely source helps Carnegie Mellon University make computers more polite
Share this post:
Researchers at Carnegie Mellon University developed an automated method for making computers more polite. And they have the ill-fated energy company Enron to thank.
How’s that?
Well, it’s not unusual to receive impolite directives or requests such as “send me the data.” But the CMU researchers found a means of taking such communications — those that use either impolite or “neutral” language — and restructuring them or adding words to make them more well-mannered.
“Send me the data,” for instance, can become “Could you please send me the data?”
The idea of turning negative statements into positive ones is something language technologists have been looking at for a while.
Those who discovered the method will present their study on politeness transfer at the Association for Computational Linguistics annual meeting, which will be held virtually beginning Sunday.
“It is extremely relevant for some applications, such as if you want to make your emails or chatbot sound more polite or if you’re writing a blog,” said Shrimai Prabhumoye, a Ph.D. student in CMU’s Language Technologies Institute (LTI).
“But we could never find the right data to perform this task.”
The unlikely source for the data came from emails exchanged by employees of Enron, the Texas-based energy company well known for a corporate fraud and corruption scandal that led to its demise in 2001.
As a result of lawsuits surrounding the scandal, half a million corporate emails became public. They subsequently have been used as a data set for a variety of research projects.
The politeness data set was analyzed to determine the frequency and distribution of words in the polite and impolite sentences. The team then developed a system for “tagging” impolite words and phrases. Then a text generator replaces each tagged item while taking care not to change the meaning of the sentence.
The system replaced first person singular pronouns such as “I,” “me” and “mine” with first person plural pronouns, such as “we”, “us” and “our.” And, instead of routinely putting “please” at the beginning of a sentence, the system began inserting it within the sentence. “Could you please send me the file?”
But how does one define politeness?
“It’s not just about using words such as ‘please’ and ‘thank you,’” Prabhumoye said. “Sometimes, it means making language a bit less direct, so that instead of saying ‘you should do X,’ the sentence becomes something like ‘let us do X.’”
For their study, the CMU research team, which also included LTI master’s students Aman Madaan, Amrith Setlur, and Tanmay Parekh, restricted their work to speakers of North American English in a formal setting.
The Air Force Research Laboratory, Office of Naval Research, National Science Foundation, Apple and NVIDIA supported the research.