Abstract:

Machines work similarly to humans due to advanced technological hy-potheses. The best example is a conversational agent that depends on advanced models in computer science. Conversational agents serve as a channel for com-munication between man and machine. Many conversational agents and designs are available on the market that work on different functions and can be applied in areas like business, medicine, agriculture, etc. The technology used for con-versational agent development is Natural Language Processing (NLP). It ad-vances the concepts of Artificial Intelligence (AI), precision, and perfection have significantly improved, and conversational agents have become the optimal choice in most organizations. Conversational agents available in travel and tour-ism areas that gather user searches and offer suitable search outcomes. Research still occurs in the conversational agent’s area to improve customer satisfaction. The research is an application that allows users to converse with humans using artificial intelligence principles. As a reference implementation, the receptionist role of a university is chosen to demonstrate the proposed solution. The typical questions encountered by the receptionist are collected with the relevant answers. Subsequently, this data is used to model a Question-Answering (QA) System to learn and answer the questions posed by the human communicator.

Among the following classifiers, Multinomial Naïve Bayes (NB), Multinomial Logistic Regression (LR), Support Vector Machine (SVM), K Nearest Neighbors (KNN), and Decision Tree (DT) are used to find the best classifier for the con-versational agent. The proposed work is to build a better model to help the stu-dents and parents, including all the academic clarifications with the help of a classifier. Various machine learning algorithms are used to test the suitability of the pro-posed system, and it has been found that multinomial Naïve Bayes is giving the best results with an accuracy of 84%. Subsequently, a conversational agent is developed to demonstrate the utility of the proposed solution with 83% accuracy. The results achieved are encouraging, and future directions are to investigate the ability of the system to work under uncertainty and incompleteness.

Keywords: Question-Answering systems, Natural Language Processing, Naïve Bayes, Support Vector Machine, Logistic Regression, K- Nearest Neighbor, Decision Tree.

Conference Published in: 7^th International Joint Conference on Advances in Computational Intelligence IJCACI 2023