Extracting U.S. common law case names (e.g., Smith v. Jones) systematically using off-the-shelf NLP software?

Question

I'd like to find a way to extract case names from U.S. courts from sentences. They usually take a predictable pattern, although I think they may be too varied to capture well with Regexs, so I was thinking about using NLP to locate them.

Here are a few examples of case names (bolded) as they might be used in partial sentences:

• In United States v. George, the Court held that...
• In re. Bankruptcy of Sir Walter Williams, III is a case from the Southern District of New York...
• Not only is Ashcroft v. Iqbal, 556 U.S. 662 (2009) incorrectly decided, it also...
• The Court's recent decision in Burwell v. Hobby Lobby Stores, No. 13-354 (U.S. Jun 30, 2014) implicates First Amendment rights...
• The case of Trans World Airlines, Inc. v. Flight Attendants was correctly decided...

I've been experimenting with off-the-shelf packages (like TextBlob for Python), which helps do things like extract noun phrases -- I just don't know how to take the next step and recognize case names as a unit.

Answer 1

How about:

((re\.).*?,.*?\b(?<=\s)(?=[a-z]))|(?!\r|\n|\.)((\s\m[A-Z][a-z]+?\M\s).*?v\.\s.*?\b[A-Z].*?[a-z]\M)(?!\s[A-Z])|Ex\sparte\s\b[A-Z].*?[a-z](?=(\.|,|;|\s))

It's imperfect in that it doesn't capture only the bolded text (it might grab a little more, but it won't match a false-positive (as it needs to find the v.)), but it's guaranteed to find all the provided examples, plus all the Ex parte cases too that I gleaned from Wiki

There are three capture groups in this regex:
1. Matches with v.
2. Matches with re.
3. Matches with Ex parte

ps: this is generic PCRE regex pattern syntax. Most program/scripting languages and many of the more advanced text editors should find matches using this.

Answer 2

Illinois Wikifier will get most of these cases for you: http://cogcomp.cs.illinois.edu/demo/wikify/?id=25