Dataset concerning: Lilobot: a cognitive conversational agent to train child helpline counselors on social support

Datacite citation style:
Grundmann, Sharon; Al Owayyed, Mohammed (2023): Dataset concerning: Lilobot: a cognitive conversational agent to train child helpline counselors on social support. Version 1. 4TU.ResearchData. dataset. https://doi.org/10.4121/7b024697-659a-47ad-95a4-0497bf52b432.v1
Other citation styles (APA, Harvard, MLA, Vancouver, Chicago, IEEE) available at Datacite
Dataset
choose version: version 2 - 2023-11-13 (latest)
version 1 - 2023-07-19

Data collected as part of an evaluation study of a BDI-based conversational agent for training counsellors of a child helpline to publish a paper, "Dataset concerning: Lilobot: a cognitive conversational agent to train child helpline counselors on social support". This dataset contains survey responses of participants with regards to two measures - counselling self-efficacy and perceived usefulness of the agent. It also contains double coding analysis and training materials. The design of the study is available at https://osf.io/hkxzc. This dataset was built on a previous dataset made for Sharon Grundmann's Master thesis: https://doi.org/10.4121/17371919

history
  • 2023-07-19 first online, published, posted
publisher
4TU.ResearchData
format
*.docx, *.Rmd, *.pdf, *.xlsx, *.csv, *.md
organizations
TU Delft, Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS), Department of Intelligent Systems

DATA

files (8)