Author

Angie Dowdell

Date of Award

3-2019

Type

Thesis

Major

Master of Science

Degree Type

Master of Science

Department

TSYS School of Computer Science

First Advisor

Rania Hodhod

Second Advisor

Randy Brou

Third Advisor

Wayne Summers

Abstract

The inclusion of human characteristics (i.e., emotions, personality) within an intelligent agent can often increase the effectiveness of information delivery and retrieval. Chat-bots offer a plethora of benefits within an eclectic range of disciplines (e.g., education, medicine, clinical and mental health). Hence, chatbots offer an effective way to observe, assess, and evaluate human communication patterns. Current research aims to develop a computational model for conversational agents with an emotional component to be applied to the army leadership training program that will allow for the examination of interpersonal skills in future research. Overall, the current research explores the application of the deep learning algorithm to the development of a generalized framework that will be based upon modeling empathetic conversation between an intelligent conversational agent (chatbot) and a human user in order to allow for higher level observation of interpersonal communication skills. Preliminary results demonstrate the promising potential of the seq2seq technique (e.g., through the use of Dialog Flow Chatbot platform) when applied to emotion-oriented conversational tasks. Both the classification and generative conversational modeling tasks demonstrate the promising potential of the current research for representing human to agent dialogue. However, this implementation may be extended by utilizing, a larger more high-quality dataset.

Share

COinS