Context-Aware Conversational Agent for a Closed Domain Task

With the ever-growing complexity in automated business processes and systems, manual navigation through the large volumes of data and hyperlinks on applications has become impractical and inefficient. Thus, conversational agents for closed domain tasks have received considerable interest in recent years. One of the fundamental characteristics of a conversational agent is its ability to interpret a statement based on the context of the entire conversation or in the context of a specific task. Keeping this view as the central focus, we study the scope of the Rasa chatbot development framework for identifying user intent and predicting the agent’s future actions. We have presented a processing pipeline that allows optimized and accelerated data generation, action path definition and evaluation of the different components to demonstrate the generalization ability of the chatbot as a whole. The proposed application demonstrates fairly good performance in identifying user intent and completing the user’s intended action.