Acquiring Semantic Knowledge for User Model Updates via Human-Agent Alignment Dialogues: Scenarios & Dialogues

Datacite citation style:
Chen, Pei-Yu (2023): Acquiring Semantic Knowledge for User Model Updates via Human-Agent Alignment Dialogues: Scenarios & Dialogues. Version 1. 4TU.ResearchData. dataset. https://doi.org/10.4121/b7a321df-640a-483d-8c32-a18fe21e7204.v1
Other citation styles (APA, Harvard, MLA, Vancouver, Chicago, IEEE) available at Datacite
Dataset

The data is the scenarios and dialogues used in the focus groups underlying the paper "Acquiring Semantic Knowledge for User Model Updates via Human-Agent Alignment Dialogues: An exploratory focus group study" by Pei-Yu Chen, Myrthe Tielman, Dirk Heylen, Catholijn Jonker, and Birna van Riemsdijk. The goal of this paper is to explore what potential users like or dislike about certain aspects of the dialogues and identify dimensions that are important for designing good alignment dialogues.


Study

We performed an exploratory focus group user study, in which we showed participants six scenarios with different variants of how we envision such alignment dialogues might look like. The scenarios and dialogues are in textual form. After they finished reading, we asked them to discuss and compare the dialogues, and then we moved on to the next scenario and discussion. The process continued until all six scenarios were discussed. For the structure of the discussion, we prepared the following questions to guide the participants:

  • Which version of the dialogue do you prefer, or which part of which dialogue do you prefer? Why?
  • Is there a certain part of the dialogue that you particularly like/not like? Why? How would you want to do it instead?
  • Which dialogue is more ‘intelligent’, as in has more capability in providing support?
  • Do you feel one dialogue is more supportive than the other?
  • After which dialogue do you think the agent would be more ‘on the same page’ as you?


Data & analysis

We transcribed the focus group sessions and analyzed the transcriptions using inductive thematic analysis with the addition of triangulation with literature.

history
  • 2023-04-13 first online, published, posted
publisher
4TU.ResearchData
format
*.pdf
funding
  • Hybrid Intelligence Center (a 10-year programme funded by the Dutch Ministry of Education, Culture and Science through the Netherlands Organisation for Scientific Research).
organizations
TU Delft, Faculty of Electrical Engineering, Mathematics and Computer Science, Department of Intelligent Systems

DATA

files (1)