Conversational Agents for Mental Health and Well-being: Discovering Design Recommendations Using Text Mining
Conversational agents are increasingly being used by the general population due to shortages in healthcare providers and specialists, and limited access to treatments. They are also used by people to deal with loneliness and lack of companionship. As these apps are increasingly replacing real humans, there is a need to explore their design features and limitations for better design of conversational apps. Using text mining and topic modeling, this study analyzed a total of 126,610 reviews about Replika, a popular and well-established conversational agent mobile app. Our results emphasized current practices for designing conversational apps while at the same time sheds the light on limitations associated with these apps. Such limitations are related to the need for better conversations and intelligent responses, the need for advanced AI chatbots, the need to avoid questionable and inappropriate content, the need for inclusive design, and the need to address some technical limitations.
Wahbeh, Abdullah; Al-Ramahi, Mohammad; El-Gayar, Omar; Elnoshokaty, Ahmed; and Nasralah, Tareq, "Conversational Agents for Mental Health and Well-being: Discovering Design Recommendations Using Text Mining" (2023). Faculty Research & Publications. 316.