NWEA - Applying Natural Language Processing based AI to Academic Assessment

NWEA - Applying Natural Language Processing based AI to Academic Assessment

Adrian Low
Noah Sperling
Evan Sterba
David Jablonski

CS 484A - Computer Science Capstone II

Description

Northwest Evaluation Association (NWEA) is a research-based not-for-profit organization that creates computer adaptive tests. Computer adaptive tests measure growth and proficiency of pre-K through 12th grade students in a variety of subjects. The tests are administered online and adjust in difficulty to match the individual student’s proficiency based on their real-time results. NWEA’s content team categorizes the questions, which are referred to as ‘items’, manually. The metadata these questions are categorized with is used to generate an appropriate test for a given student, subject and grade level. The manual categorization of each item is time consuming and is prone to human error. To save the content team time and create more consistent categorization, the team developed an internal tool using natural language processing to investigate the potential to automate the metadata creation process. The hope is that NWEA can use this project as a proof-of-concept to build a universal tool in the future.