Discourse Representation Framework for Consumer NLP

With the ultimate goal of enhancing Nexxt Intelligence’s market research SaaS platform, inca, this project will create a flexible framework for facilitating conversations with consumers using an agent powered by natural language processing and deep learning models. This ‘discourse representation framework’ will be capable of engaging with consumers in fluent English, and is designed to ask sensible and specific questions which are pertinent to ad-hoc research objectives. The agent will do so by adhering to a ‘discourse policy’, which will be a non-technical conversational design document which can be easily created by market researchers for their research goals. The framework will simultaneously extract structured information from user utterances, using a combination of deep learning and rule-based heuristics, and will navigate through the conversation based on the information it is able to extract.

Yuxiao Sun
Faculty Supervisor: 
Shurui Zhou
Partner University: