Take each section of a country's law and convert it into prolog clauses. Queries can then be run on the legal engine. For defendants it gives you insights on how to build a case and for prosecution it identifies relevant sections and evidence that needs to be provided to have a successful conviction.
The same can be applied to divorce, patents, property etc.
My idea was to use an Erlang map reduce system to help fan out the queries which are dispatched to an underlying Prolog knowledge base (Erlang supports something like Channels/Ports).
I have a bias towards ideas which have a social impact.
The business case is: In a country like India there are over 20 million pending cases in courts.
Imagine both sides of lawyers and the judge all having access to a system like this - cases could be resolved a lot faster and time spent building defense/prosecution would be a significantly smaller.
Implemented right, this could somewhat level the playing field and allow poorer people have access to some sort of legal advice, which today they would not be able to afford. Monetizing the system could be charging for queries as you probe deeper and deeper into the system/advertising for lawyers.
I think commoditizing law has immense potential and should have a very large business potential.
I always wanted to implement and Idea like this as Open Source, but here in India nobody would fund ideas like this. I'm putting it out there as I believe it's time has come.
Take Argentina, a case I know far too well. As INC magazine pointed out, the tax rate on businesses goes up to 108% of your profit. (< http://www.inc.com/magazine/201106/doing-business-in-argenti... >). How's that possible? Here's how: the government purposely passes endless contradictory laws thus ensuring you are always breaking the law, just to survive. (A business can't pay 108% of its money in taxes and still survive!). Therefore, you live in a state of somehow doing something illegal. The result? If the government doesn't like you, they find the illegal thing you're doing (that you have to do, just to survive, since the laws are contradictory and unreasonable, like that 108%), and then punish you for "breaking" the law.
Welcome to the jungle, we've got fun and games ;)
Conclusion: If you step into the third/developing world, the key isn't what the law actually says -- but _how sh*t gets done in practice_.
I think the base idea is very inspired, but I'd avoid prescribing a technical solution at the beginning and just propose different approaches. The general goal of giving lawyers a formula for inputting a case with a certain grammar and format and getting a useful output of laws and precedence seems amazingly useful.