The effect of emotional interaction on user satisfaction in chatbots
Keywords
Loading...
Authors
Issue Date
2023-01-27
Language
en
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Chatbots have been making their way into our lives in a quickly increasing number
throughout recent years. Used in a lot of domains, such as health and management, the
need for keeping a high standard is exceptionally present. As the chatbot is capable of
substituting humans, we can ask the question if having a chatbot more similar to an
emotional being (such as a human) is useful. This leads to the research question, where
we ask if emotional interaction influences user satisfaction in conversation with a
chatbot. Known methods of conveying emotion are implemented in the chatbot to get it
as close to an emotional interaction as possible. Multiple factors are tested by means of
a questionnaire to get insights if they are influenced by emotional interaction or not.
This is done by creating a chatbot that is not acting emotionally and one that is and
having two separate groups interact with them. These chatbots are vacation
recommendation systems. Then we test if these factors are correlated with user
satisfaction to conclude our research.
With the data we received, we performed a t-test to find a significant difference between
the influence of the different chatbots on the factors. We used Pearson’s correlation
coefficient to find out if the factors correlate with user satisfaction. Even with limited
data, we are able to make solid inferences about the hypotheses. Only ease of use was
found to not be influenced by the emotional chatbot against the non-emotional chatbot.
All the other factors, which are, perceived trust, honesty, predictability, and capability,
are all found to be significantly higher when interacting with an emotional chatbot as
opposed to a non-emotional chatbot. Combining this with the findings that all factors
are positively correlated with user satisfaction, we can say that emotional interaction
does indeed influence user satisfaction in chatbot conversation. This conclusion is
however to be taken lightly, as this research lacks scientific robustness.
Description
Citation
Faculty
Faculteit der Sociale Wetenschappen