“Alexa, what do you do for fun?” Characterizing playful requests with virtual assistants

Digital assistants these types of as Siri or Alexa are an inseparable section of our life today. Nonetheless, they nonetheless battle with some utterances. For instance, virtual assistants cannot reply to humorous requests, like a person asking to acquire a mind. A recent paper implies employing scientific theories of humor to characterize these requests.

Image credit history: Y2kcrazyjoker4 by using Wikimedia (CC BY-SA four.)

A taxonomy of playful utterance types, pushed by true-planet info from Alexa, was launched. The scientists focused on a distinct section of taxonomy, the personification of the virtual assistant as a human, for a person analyze. The design instantly identified personifying utterances. Guide solutions had been created if the design detected utterances that cannot be addressed by predefined solutions. It was revealed that the person working experience is enhanced when the virtual assistant understands their playful utterances and responds appropriately.

Digital assistants these types of as Amazon’s Alexa, Apple’s Siri, Google Home, and Microsoft’s Cortana, are turning out to be ubiquitous in our each day life and efficiently aid people in a variety of each day tasks, these types of as building mobile phone calls or enjoying new music. Yet, they nonetheless battle with playful utterances, which are not meant to be interpreted actually. Examples include jokes or absurd requests or issues these types of as, “Are you worried of the dark?”, “Who permit the puppies out?”, or “Order a zillion gummy bears”. These days, virtual assistants often return irrelevant solutions to these types of utterances, except for hard-coded ones addressed by canned replies.
To deal with the obstacle of instantly detecting playful utterances, we initial characterize the diverse types of playful human-virtual assistant conversation. We introduce a taxonomy of playful requests rooted in theories of humor and refined by analyzing true-planet targeted traffic from Alexa. We then aim on 1 node, personification, where people refer to the virtual assistant as a individual (“What do you do for fun?”). Our conjecture is that being familiar with these types of utterances will improve person working experience with virtual assistants. We performed a Wizard-of-Oz person analyze and confirmed that endowing virtual assistant s with the means to recognize humorous opportunities in fact has the prospective to raise person pleasure. We hope this function will add to the being familiar with of the landscape of the problem and inspire novel strategies and strategies toward the eyesight of supplying virtual assistants a perception of humor.

Investigation paper: Shani, C., Libov, A., Tolmach, S., Lewin-Eytan, L., Maarek, Y., and Shahaf, D., “”Alexa, what do you do for fun?” Characterizing playful requests with virtual assistants”, 2021. Link: https://arxiv.org/abs/2105.05571