Skip to main content

Intelligent Tutoring System for Non-Native, Low-Literacy Individuals

Francisco IacobelliPI: Francisco Iacobelli

Visiting Associate Professor of Computer Science - Northwestern University
Associate Professor of Computer Science - Northeastern Illinois University

Faculty Profile

Conversational systems (e.g., Siri, Alexa, Google assistant) are increasingly used for educational purposes but biases prevent their effective use by those who need them the most: individuals with low levels of literacy and educational attainment. 

An important issue in large scale learning models is biases in data and design decisions. The data selected to build systems on, and the way that data is presented impacts who can understand a system and inversely, who can be understood by the system; and ultimately the learning gains of the target user. This is especially true among minority groups, wherein a model may be trained on, and present data in a way that is not conducive to learning for members of that group. In the case of low-literacy minority individuals, system designers may attribute issues with a system’s performance to the user’s inability to articulate commands correctly or comprehend instructions, while users would attribute performance issues to the system’s inability to communicate with them. 
 
This project is a case study that will consist of the design of an intelligent tutoring system for low-literacy Latina breast cancer survivors. The system will train the women on breast cancer survivorship skills. Developing a system to accomplish this specific task is also intended to provide a framework for developing future learning models and conversational systems that respect cultural norms and linguistic practices of minority populations. This project also intends to provide a standard framework for end-to-end design of intelligent systems that target minority populations, at scale.

Back to top