The Use of Paraphrase Identification in the Retrieval of Appropriate Responses for Script Based Conversational Agents

This paper presents an approach to creating intelligent conversational agents that are capable of returning appropriate responses to natural language input. Our approach consists of using a supervised learning algorithm in combination with different NLP algorithms in training the system to identify paraphrases of the user’s question stored in a database. When tested on a data set consisting of questions and answers for a current conversational agent project, our approach returned an accuracy score of 79.15%, a precision score of 77.58%and a recall score of 78.01%.