This projects aims at creating an intelligent tutor that can develop an interactive and intelligent dialogue with the student.
A dialog system or conversational agent (CA) is a computer system intended to converse with a human, with a coherent structure. Dialog systems have employed text, speech, graphics, haptics, gestures and other modes for communication on both the input and output channel. A traditional dialogue system consists of the following components:
A natural language understanding (NLU) module, which receives utterances of free texts (typed or spoken), and maps them into a structured semantic frame; usually there are three key tasks in such a NLU module: domain classification, intent determination and slot filling. Typically, there exist two kinds of recipes: single-task learning with a concatenated approach and multi-task learning with a joint approach.
A dialogue manager (DM), which consists of a state tracker and a policy learner: the state tracker offers the ability to access the external database or knowledge base, tracks the evolving state of the dialogue, and constructs the state estimation, whereas the policy learner takes the state estimation as input and chooses a dialogue action.
A natural language generation (NLG) module, which translates the structured dialogue action representation into a natural language form