MOODs: Identifying Outlier Passages and Texts in the Legislative Process
Dr. Peter Organisciak
RMIS Faculty, University of Denver
Summer 2018 - Summer 2019
MOODs is an adjacent project similar to the IMLS grant-funded SaDDL project. While SaDDL looks at data from the HathiTrust, the MOODs project analyzes congressional bills. The goal of this project is to develop a way to help readers judge when and where pieces of legislature warrant closer reading. The system, called MOODs, aims to find textual Misfits, Omnibuses, and Odd Ducks.
The project uses different methods to analyze legislature. First, the project uses topic modeling to create a context for what is "normal" in a bill. Based on that model, differences will be measured in two ways: measuring the fit between the model and the text and measuring how well one set of documents represents another. The project then considers how texts are distrbuted across those topic models and ends by developing a measure to indicate bills that are outliers.
The MOODs project focuses on three research questions:
What are the most novel federal bills currently?
What are the most topically-diverse federal bills currently?
What are the most thematic or thematically-unexpected passages in relation to a bill's overall content?
The overall goal of MOODs is to allow better understanding of the political process in addition to helping people understand legislative texts in general.