Conversational Modeling and Dialogue Generation for Consumer Understanding with Neural Techniques

Market research is fundamentally about understanding humans as consumers, in all their complexity and nuance. This fellowship will apply natural language processing techniques to the problem of representing consumer language, and generating conversational prompts intended to elicit opinions and underlying motivations which are pertinent to ad-hoc market research objectives. In partnership with Nexxt Intelligence Inc., this project will leverage recent developments in machine learning to construct a deep learning model capable of categorizing consumer conversations into meaningful categories and relationships which are of interest to market researchers. This model will then be applied to downstream tasks, building on existing IP (including a consumer language dataset, language models, and a computational conversational framework) toward the ultimate goal of creating an end-to-end digital system which empowers market researchers to design and conduct conversational experiences which engage with consumers through the use of fluent questions and prompts, which are not only sensible but also specific to both the market researchers’ overall objectives as well as the consumers’ utterances. The knowledge gained through this fellowship will be disseminated to the scientific community through publications in peer-reviewed journal(s) and/or conference(s), and the resulting models will be published with an open source license for anyone to use.

Faculty Supervisor:

Jimmy Lin

Student:

Partner:

Nexxt Intelligence

Discipline:

Computer science

Sector:

Professional, scientific and technical services

University:

University of Waterloo

Program: