Somehow we slipped past the 250 million fact mark without any fanfare!
We keep a count of the number of facts because it's one of the main contributors to our question-answering capability (by no means the only factor, though). You might be wondering what we mean by a 'fact'. Well, we're talking in this case about individual pieces of knowledge asserted either by users of our site (you!), or by the automated processes that our mining team run. Facts can be grouped into three broad categories:
- factual
[london] [is geographically located within] [the united kingdom] - common-sense
[city] [is a subclass of] [settlement] - lexical
["london"] [can denote] [london]
An important point to remember is that all these facts are stored in a single, unified representation, not scattered about in a variety of different database tables.
Of course, one of the things that sets True Knowledge apart, is its ability to use knowledge to infer new knowledge. So, if we have a fact in the Knowledge Base that London is in the UK, and another fact that the UK is in Europe, we don't also need a third fact that London is in Europe—that can be inferred from the first two facts. Just because we have 250 million facts doesn't mean that True Knowledge is limited to knowing 250 million things—it can infer far, far more.
We reached 100 million in May 2008, and the 200 million mark was passed at the beginning of September last year. Facts aren't the whole story, but they go a long way—roll on half a billion!
a couple of questions..
do you have an rdfs or owl ontology?
can you make any data available as rdf?
would you be open to converting or exposing your knowledge base as linked data?
could your technologies be refactored to work over rdf+owl?
ty for any reply, nathan
Posted by: Webr3 | 16 April 2010 at 09:44 PM
Well done on passing the quarter of a billion mark, what an achievement!
Posted by: Glasgow bathroom | 21 April 2010 at 12:10 AM
That is amazing. I understand how hard it is to generate that much information.
Posted by: Trey in Bose Idaho | 28 April 2010 at 06:13 PM
See http://blog.trueknowledge.com/2010/02/at-true-knowledge-we-are-passionate-about-linked-data-and-making-our-api-as-accessible-as-possible-we-know-there-is-a-huge-c.html for discussion of linked data.
On the rest we believe we could build a sparql endpoint over our platform but none currently exists.
Posted by: William Tunstall-Pedoe | 27 May 2010 at 10:06 PM