Artificial intelligence for human-computer interaction

• Main Author: Low, Shi Min
• Other Authors: Erik Cambria
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2022
• Subjects: Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/156670
• Institution: Nanyang Technological University

In recent years, conversational agents have promised new opportunities to shape human-computer interactions. These dialogue systems are a form of artificial intelligence (AI) that accepts a human utterance as input and responds with an appropriate computed answer as the output. The successful implementation of such dialogue systems may potentially reduce human resources and increase operational efficiency. Therefore, human-computer dialogue systems have gained enormous interest in the service sector as more services are choosing to integrate them into their systems. To provide better human-computer interaction, the current implementation of dialogue systems should shift away from generic and robot-like responses. Therefore, this study aims to improve the current implementation and explore the usage of emojis in dialogue systems to further enhance human-computer interaction. In addition to the popular Generative Pre-trained Transformer 3 (GPT-3) model, the approach taken in this study further fine-tunes this model to obtain more desirable responses. This is followed by the generation of emojis in a dialogue system’s response to amplify a more human-like response. Lastly, this is combined to create a dialogue system capable of performing dialogues that mimics a human. Overall, this study has shown promising results to enhance human-computer interaction.