The goal of the project was to create a coreference resolution model that could identify entities being referred to in a social bot’s user utterance using information found in our discourse model. This was addressed by implementing both a rule-based, neural, and ensemble model, and comparing how these performed on a set of pronominal types: it, they, that, she, he, her, him, his, and hers. To enhance the models, the team harnessed the information found in their discourse model that tracked and held information of which entities are currently in focus in the conversation, independently of whether they are realized by a pronoun or a definite referring expression or a proper noun, and independently of whether they are introduced into the discourse by the user or the system. The team’s objective was to improve the system’s ability to acknowledge the entity being referred to by the user and respond with more salience.
Student team: Jeshwanth Bheemanpally, Phillip Lee, Cecilia Li, Angela Ramirez, Eduardo Zamora
Project Mentor: Professor Marilyn Walker & Dr. Adwait Ratnaparkhi, UCSC
Amazon Alexa Prize Discourse Model & Coreference Resolution (PDF)