Rebecca J. Passonneau
Professor of Computer Science
An automated process to score student summaries of documents is a project Rebecca Passonneau has already been working on, and TLT will be helping to expand that work in her 2017-18 TLT Faculty Fellows project.
Currently, Passonneau uses exemplar summaries to rate the student works against — seeing where they align and where there are outlier ideas that could be either stellar points not previously made or poor points unsupported in the documents. This process speeds up feedback for students, but the process which relies on high-quality, written examples is not scaleable. She will be working with TLT to develop an application to turn her research into a process that could work on a larger scale and in more contexts.
A potential means to instill stronger language skills would be to develop more sophisticated digital learning environments that allow students to interact with their curricula more consistently in reading, writing, and revision exercises, and that facilitate teachers’ ability to monitor and promote students’ reading and writing skills through timely analysis of their written work. The software for content analysis of summaries that we have developed can identify which ideas students have mastered, versus those they still struggle with, given a small number of models written by proficient individuals.
Passonneau envisions additional steps for her fellowship, including holding a workshop showing how the system could be adapted for other use cases, such as formative assessment of student writing, and finding ways to extend the algorithmic work from feedback on summaries to feedback on essays, which have similar but also differing qualities.
Drew Wham (lead)