Collaborating with a Text-Based Chatbot: An Exploration of Real-World Collaboration Strategies Enacted during Human-Chatbot Interactions

要旨

A central problem for chatbots in the customer care domain revolves around how people collaborate with the agent to achieve their own situated goals. The majority of the previous research, however, relied on experiments within artificial settings, rather than on observation of real-world interactions. Moreover, such research mostly analyzed users’ responses to communication breakdowns, rather than the wider collaboration strategies utilized during a conversation. In this paper, we qualitatively analyzed 12,477 real-world exchanges with a task-based chatbot using a Grounded Theory approach as a rigorous coding method to analyze the data. We identified two main aspects of collaboration, behavioral and conversational, and for each aspect we highlighted the different strategies that users perform to “work together” with the agent. These strategies may be utilized from the very beginning of the conversation or in response to misunderstandings in the course of ongoing interactions and may show different evolving dynamics.

著者
Amon Rapp
University of Torino, Torino, Italy
Arianna Boldi
University of Torino, Torino, ITALY, Italy
Lorenzo Curti
University of Torino, Torino, Italy
Alessandro Perrucci
TIM S.p.A., Torino, Italy
Rossana Simeoni
TIM S.p.A., Torino, Italy
論文URL

https://doi.org/10.1145/3544548.3580995

動画

会議: CHI 2023

The ACM CHI Conference on Human Factors in Computing Systems (https://chi2023.acm.org/)

セッション: Conversational Agents

Hall F
6 件の発表
2023-04-26 20:10:00
2023-04-26 21:35:00