Towards Artificial Intelligent Conversational Agents : Expository Communication Styles for Building Chatbots Using Natural Language Processing

Abstract

Expository writing is one of the most common styles of oral and written communication. The four types of expository writing are expository, persuasive, narrative, and descriptive. Despite that it is commonly taught in secondary schools in the U.S., there have not been previous studies on how expository writing can lead to useful features and metrics for natural language understanding. This thesis found that there are significant differences in the semantic and syntactic features among corpora representing the four styles. These results inspired a new communication style categorization for chatbots dialog designs. It also suggests a new approach for conversational agent architecture.