IBM Research has launched a companion service to its Many Eyes data visualization service called Many Bills, which provides a visual interface for federal legislation. The company says the idea behind the project is to make it easier to understand complicated bills that run into the hundreds of pages and often have unrelated amendments tacked onto them by Congress. IBM’s algorithms parse each bill (the service currently just has legislation from 2009) and then color-codes sections of them based on topics. The service even includes a “confidence score” so that readers can decide for themselves whether its categorization process got it right or not.
The visual interface takes a little getting used to at first. Each bill is laid out in a long, vertical stream with different colors used to identify different categories, but the discrete parts of the bill are difficult to see because there are hundreds of them, so they are relatively tiny. When you click on one, it expands to a larger view, with annotations about the section and what it relates to, the different terminology used, etc. It would be nice if there were another way to go through the pages, such as a slideshow, as the default view with hundreds of tiny colored sections is a little hard to navigate.
The venture seems like a positive one, however, and has been applauded by the Sunlight Foundation – a non-profit agency that promotes transparency in government — which said that it sees “a lot of potential for this project.” TechPresident, a blog affiliated with the Personal Democracy Forum, said that while the service is far from perfect, it’s valuable because it helps — even in a small way — to make the federal legislative process more understandable. You can follow the project’s progress via the Many Bills Twitter account.
Although it’s not designed specifically for federal legislation, Many Bills is somewhat similar to DocumentCloud, a project started by Aron Pilhofer of the New York Times’ interactive team and funded by a grant from the Knight Foundation, a non-profit that invests in journalism. Organizations can upload documents and the service scans them and extracts keywords so that they can be searched and filtered. Both DocumentCloud and the Many Bills project are attempts to use machine learning to help people make sense of complicated documents, and heaven knows we can use all the help we can get.
Related content from GigaOM Pro (sub req’d):