We are building a natural language interface to databases tool. I figure that forms, key-word and hyper-link interfaces have their limitations. Our orientation is mostly scientific/technical, but we figure that if we can deliver a solid, usable solution, there will be many, many uses.
But of course there are significant challenges. :-)
take a gander at our demo clip for VLDB.
http://www.youtube.com/watch?v=fWio8bHq4wQ
That was our first startup idea with Organon Technologies. We realized that it was going to take us a couple years to build it out, so we switched to streamfocus.com instead. Good luck!
Well I will tell you that it was quite an effort. There are lots of moving parts under the hood - a theorem prover, search mechanisms, spell checker, etc.
It took several years to get it to the current beta. The nice thing I have tenure as an associate professor in Sweden so I can grind on this problem with little interference and a quirk in Swedish law lets Anaphoric Systems (our company) own it. The sacrifice has been that my publication rate slowed down. But honestly I don't really care (that much).
Still I have been told that it is hard to sell a generic database interface tool. I have had lots of advice to specialize in one vertical or another. But which one to pick?
My own dream would be to sell it on the database tools market as a general solution.
Still if our Anaphoric Systems start up fails to get traction, we will probably just open source the whole project -- and I will get back to publishing on areas like causality, probability logic, etc.
I just looked at your demo - nice job! If you're interested, our application streamfocus.com might be a great platform for you - perhaps we could collaborate?
Absolutely! I just looked at your website and I think there is an area of possible collaboration. Let me study it a bit more. But a quick question, to what extent are you using relational databases to represent base level facts in your system?
Great! We're using Mysql for our database, so I think it would be the perfect platform for seeing what you can do with real world company data and hopefully adding excellent value to our service - a win win. (I've just emailed you to get on the beta)
Business model idea: We could let people use your search free for the first month, and then let them decide if they want to keep using it for an additional amount of money on top of our base subscription per month.
