ChatGPT based ChatBot Application

Since the invention of ChatGPT, there have been several applications demonstrating how ChatGPT can be used to solve real world problems. This paper focuses on implementing one such application, where we design ChatBot using ChatGPT API. To carry out the experiments, we choose dataset from a specific domain, and store the data in graph oriented database. Sample training dataset consisting of English and its equivalent Cypher queries were designed for prompt tuning ChatGPT on the local Knowledge-graph. This training dataset helps ChatGPT understand the mapping between the English and their equivalent Cypher queries. Once the mapping is learnt, it becomes easier for ChatGPT to encode or represent a real time (previously unseen) English query fed by the user into its equivalent Cypher query format. In this paper, we demonstrate our results on sample Rope-related dataset scrapped from internet, and conclude with limitations and future work.