Real-Time Attention-Based Conversational Agent

Neural Machine Translation (NMT) is a prominent natural language processing technique that is being used to develop conversational AI technology. Traditional industrial chatbots often rely on scripted responses and lack the ability to provide real-time data-driven interactions. Developing advanced AI chatbots like Google Bard or ChatGPT are out of reach for smaller or medium-sized organizations due to their scale and associated costs. Chatbots are majorly restricted by the data on which they were trained on and have no knowledge of current events. This research project intends to research and develop an approach that enables chatbots to provide live and up-to-date information in their responses and can be developed in minimalistic costs so that even smaller or medium-level organizations can afford to provide interactive AI chatbots. We experiment with various techniques in terms of the type of data being used to harness live capabilities. We focus on optimizing the hyperparameters required for building a conversational AI agent and leverage open-source technologies to minimize costs. To ensure flexibility and affordability, we adopt a microservice architecture that combines Attention based NMT models and Transformer Models with live API services features, leveraging the RASA actions API. This approach allows us to develop a prototype of an advanced chatbot that goes beyond the traditional scripted responses by providing real-time information to users and being affordable to develop.