Using Model Transformation to Generate Chatbots for Web APIs

Abstract

Chatbots provide a novel form of user interface, and Web API (Application Programming Interface) technology is rapidly expanding to encompass a wide range of applications. The integration of Chatbots with APIs facilitates access to a wide range of data and functionalities; however, existing tools for the development of a conversational Chatbot user interface can be exceedingly complex. This paper presents a semi-automatic system based on the concept of MDE (Model-Driven Engineering) by which to generate a Chatbot user interface for a specific Web API in accordance with the proposed BotSwagger specifications. The devised BOTEN system allows developers to integrate APIs with Chatbots seamlessly. It also enables end users to input parameters, access APIs, and obtain comprehensible responses promptly in a conversational way. In experiments, the proposed system proved highly effective in integrating APIs with Chatbots. It makes it possible for users to input required parameters sequentially via an interactive dialogue to use Web APIs. The proposed system simplifies the process of creating Chatbots with API services, while largely eliminating the need for API-specific knowledge. Besides, the proposed LLMbased coordination mechanism (LLM: Large Language Model) also facilitates the creation of large-scale applications comprising multiple Chatbots.