- Computational Social Science
- Critical Data Studies
- Data Science
- Economics and Information
- Education Technology
- Ethics, Law and Policy
- Human-Computer Interaction
- Human-Robot Interaction
- Incentives and Computation
- Infrastructure Studies
- Interface Design and Ubiquitous Computing
- Natural Language Processing
- Network Science
- Social Computing and Computer-supported Cooperative Work
- Technology and Equity
The Information Science Breakfast Series is a seminar-style meeting held one morning a week. The Series provides a venue to share, discover, and discuss research happening in Information Science. It is intended to be a less formal environment; and practice talks, works in progress, and elicitations for feedback on early-stage ideas are all welcome forms of presentation.
The guest speaker for this week will be Elijah Mayfield.
Talk Title: “Analyzing Authoritative Conversational Language with Machine Learning”.
Abstract: We constantly interpret social cues in our daily life - from members of other cultures, in new environments, and discussing complex topics. As common as these interactions are, it is no surprise that judging the authoritativeness of others is intuitive, almost automatic, for humans. For computational modeling, however, these cues are not so easy. My research builds sociolinguistic insight into machine learning, streamlining their theories when necessary and automating them when possible, to allow for patterns of authoritative behavior to emerge from large collections of conversational data.
This talk demonstrates that authoritative language in conversation can be defined in a reliable way; that those cues can be reliably recognized using machine learning, matching human accuracy in judgment of authority and annotation of social behavior; and that the output of that prediction correlates with diverse outcomes, such as interpersonal trust, task success, and cross-cultural communication effectiveness. These results range across conversational domains, from controlled lab studies to the high stakes domains of healthcare and education.
Bio: Elijah Mayfield is a Ph.D. student at the Language Technologies Institute at Carnegie Mellon University, and a recipient of the 2011 Siebel Scholarship and the 2013 IBM Ph.D. Fellowship. His research focuses on using machine learning to better understand and recognize social cues in language, with an emphasis on authoritativeness, expertise, and empowerment. In practice, he is using these insights to improve understanding of real-world problems in education and healthcare. Elijah is also the founder of LightSIDE Labs, a startup in Pittsburgh, PA building platforms for automated assessment of student-written texts using machine learning. In 2012, LightSIDE was invited to participate in the ASAP Competition, held by the Hewlett Foundation and Kaggle.com, where those vendors demonstrated that the state-of-the-art in automated assessment matches human reliability. His work with LightSIDE has been highlighted across a variety of news sources including NPR Morning Edition, Education Week, and Science.