Identifying intentions in conversational tools: a systematic mapping
Resumo
Context: The utilization of chatbots/conversational tools for various tasks has increased following the advent of ChatGPT, thereby opening up greater possibilities in human-computer interaction. Problem: Despite extensive investigation, fully mastering the linguistic and computational challenges in identifying intentions in dialogue acts remains elusive. Solution: To advance the field, a comprehensive review and synthesis of existing research is essential, with the aim of achieving greater precision in intention identification for improved user outcomes in tool utilization. IS Theory: The theoretical foundation is based on intentions and dialogue acts. Method: A systematic literature review was conducted on ACM DL, SCOPUS, Science Direct, and Springer Link databases, where 201 studies were retrieved. Among these, 39 were selected for a comprehensive review. The analysis of the results was conducted using a descriptive approach. Summary of Results: Eighteen different techniques/architectures/models and thirty-three datasets emerged from the studies, exploring ten different business domains. Contributions and Impact in the IS Area: The use of chatbots can impact various domains, including education, customer service, hospitality, information systems, among others. Understanding and enhancing key elements in dialogue comprehension can maximize the effectiveness of human-computer interaction, leading to improved outcomes and contributing to advancements in these tools.
