Feb 13, 2017
Jolley Hall, Room 309
Assessing Text Intelligibility with Discourse Structure
Junyi Jessy Li
Department of Computer and Information Science
University of Pennsylvania
One major aspect of communicating through text is to organize information into an understandable and well-structured discourse. This gives rise to two challenges in text processing: how to understand information organization and how to use such knowledge to characterize text intelligibility. In this talk, I will present two main aspects of my research that tackle these questions. First, I will discuss specificity, i.e., general and specific content in text. I will present an efficient semi-supervised system for predicting sentence specificity, and also talk about the close ties of text specificity to discourse structure and text comprehension. Second, I will discuss information organization factors that impact cross-lingual communication, in particular in the context of Chinese-English translation.
Junyi Jessy Li is a Ph.D. candidate in the Department of Computer and Information Science at the University of Pennsylvania. Her research interests are in natural language processing with focus on computational discourse. Her Ph.D. work has explored discourse structure in monolingual and cross-lingual context, including text specificity, content organization and discourse relations. She participated in the 2016 AAAI/SIGAI Doctoral Consortium and is a former intern at Yahoo! Research (2015) and Microsoft Research (2013). Her paper describing aspects of her Yahoo! internship was nominated for the best paper award at SIGDIAL 2016.