True Knowledge Works on a Wikipedia for Facts

ScreenshotIt’s sometimes said that the Internet contains all knowledge. The problem with that theory is twofold: finding the knowledge and making it useful. Search engines help with the first issue, but not with the second; making knowledge useful & understandable to both people and machines is the longstanding dream of the semantic web.

Now startup True Knowledge is trying to fulfill that dream using a combination of their own technology for storing facts and relations and the sort of wide-open crowdsourcing that made Wikipedia great. On their site, you can ask questions such as “How old is John McCain?” or “What is the capital of North Dakota?” and get back an answer based on their database of nearly 100 million facts, along with an explanation of how the answer was known or deduced. In addition to simply regurgitating facts, the system is also capable of combining them together using a rules-based engine.

In addition to asking it questions (which will reveal to you fairly quickly that 100 million facts really isn’t very many), you can also teach the system new things. This is done through a guided interface that will ask you questions to narrow down each term of a new fact. For example, I was able to tell it that I was the editor of Factsheet Five from 1982 to 1992, but before I could put in that simple-looking fact I had to explain what Factsheet Five was, what a magazine was, and what “is the editor of” meant, as well as locating each of those terms as well as I could within their growing ontology. All told, the process took perhaps 30 minutes – it was straightforward and the questions were well-phrased, but a bit tedious.

In order to assert some quality control on the facts being entered, True Knowledge also allows you to endorse or contradict any fact you run across – with the obvious hope that the “wisdom of crowds” will lead to a high-quality factbase. This sounds dangerously naive, but of course we can remember when people said the same thing about the writing and editing process at Wikipedia (which, despite a few refinements, still largely depends on the kindness of strangers).

At the moment, True Knowledge is mainly an interesting toy (though for some reason entering new facts has the potential to be an addictive timesink). What makes it more interesting to web workers is that they’re also planning on exposing facts and relations as APIs (indeed, some are already available that way). So potentially, if they can overcome the obstacles, they could serve as a knowledge repository for any other web application that could use such a thing. One could ultimately imagine them serving as a sales tax rate service, a way to verify mailing addresses, or a nutrition lookup service, for example.

True Knowledge is currently in closed beta; you can enter your email address and you’ll get a login as new slots are opened up. It’s a project worth keeping an eye on.

Comments have been disabled for this post