Luke Rosiak, a projects reporter for the Washington Times, has created an amazing online resources with his IRS Nonprofit Form 990 search tool.
The Form 990 is the tax return that nonprofits are required to file. These have been publicly available in PDF form for quite awhile, but generally those PDFs are not searchable.
What Rosiak has done is create a tool–for which he has released all the code as Open Source on Github–that OCRs all of those Form 990s and then organizes that information based on the fields from which it is capture. This allows interesting uses such as searching for all nonprofits that featured a given person as the contact, or all nonprofits located at a specific address.
For example, you can quickly search to see which nonprofits Ingrid Newkirk is listed as the principal officers. Very nice.
Rosiak’s database apparently includes OCRed versions of every Form 990 filed since 1999. This is a very nice example of the potential for open data projects.