João Sedoc recently finished his PhD in Computer and Information Science at Penn, and is spending a year at Johns Hopkins, after which he'll start as a faculty member in NYU's Stern School of Business. His CV lists his research interests as "Natural Language Understanding, Conversational Agents, Hierarchical Models, Deep Learning, Spectral Clustering, Spectral Estimation of Hidden Markov Models, Time Series Analysis", but his background is even broader than that list suggests.

João got a BA in Math and a BSE in Computer Science from Penn in 1999. He then worked in the securities industry until 2013, specializing in algorithms for high-frequency trading (see his CV for the details). He entered Penn's PhD program in 2013, where he focused on the development of improved conversational agents. As his research statement explains,

"I am pursuing two primary areas of research. The first focuses on developing novel deep learning architectures to build chatbots with temporal and stylistic coherence and the second develops methods for the systematic evaluation and comparison of novel chatbot models. I am interested in creating new multiscale models for chatbots which can be applied to practical areas of interest like conversational surveys for studies and healthcare, entertainment (particularly, gaming), and education. However, prior to possible applications, models need to be able to maintain topical coherence and persona, as well as match register."

His most recent publication is "Learning Word Ratings for Empathy and Distress from Document-Level User Responses".

To join us for dinner on Friday 1/31 in Memorial Tower 300 at 6:00, please sign up at.https://ware.house.upenn.edu/content/dwip-friday-131-jo%C3%A3o-sedoc

   Mark Liberman
   Faculty Director, Ware College House