Slashdot Mirror

Here is a link to the database and the question
http://www.trueknowledge.com/q/what_was_the_most_boring_day_in_history

The 'scientists' are actually employees of trueknowledge.com - an AI question answering start-up. The experiment was a fun bit of work that dropped out of compiling hundreds of millions of machine understandable facts about the world. The original story is here: http://blog.trueknowledge.com/2010/11/most-boring-day-in-history.html

How many wheels on a tricycle?
TrueKnowledge.com: 2
Fail.

http://www.trueknowledge.com/q/how_many_wheels_on_a_tricycle

No, wolframalpha is a math search/AI thingy. You're thinking of TrueKnowledge. It's pretty cool, but it also hideously ugly. Which is a shame, because they had a pretty attractive design when it was a closed beta a few months ago.

I was thinking more along the lines of True Knowledge, this project has been in private beta for about 2 years and just in the last 6 months was opened up for public beta testing. What it has so far is limited, but it is growing steadily.

Disclaimer: I do not work for True Knowledge, Google, or Wolfram Alpha, I have just been helping out with beta testing true knowledge since it was a private beta system...

Hopefully some companies, like True Knowledge ( http://www.trueknowledge.com/ ) will take off. It's beta at the moment and not perfect by any means but it's coming a long quite well (may start tinkering with their APIs myself) and it would be nice to see another real contender come into the search market and one that's no fro the same tiny section of the globe.

We do need more competition. Understandably when you competitors are Google and Microsoft, that's enough to put off a lot of people.

This seems very similar to True Knowledge, which has been in Beta for ages, and not as other people suggest, Google

The article forgot to say that Steven talk
at Harvard tomorrow and that the talk is available over webcast.

See http://cyber.law.harvard.edu/events/2009/04/wolfram

While we are waiting for alpha you might like
to play 2 other knowleadge engines like:
http://start.csail.mit.edu/
and
http://quizbot.trueknowledge.com/

PS. Also check out my news site
http://crowdnews.eu/

There is already such a search engine that has been in beta for a while. Go here : http://www.trueknowledge.com/ An earlier version (quizbot) has been available for ages (although not nearly as good as the current trueknowledge beta): http://quizbot.trueknowledge.com/

There is already such a search engine that has been in beta for a while. Go here : http://www.trueknowledge.com/ An earlier version (quizbot) has been available for ages (although not nearly as good as the current trueknowledge beta): http://quizbot.trueknowledge.com/

quizbot at
http://quizbot.trueknowledge.com/

says 206

I asked "Do human bodies have bones" and got a yes.
Wolfram alpha might be more complex, but as long it is not online we can't be sure.

Quizbot is online now and is able to answer more question every day.

See
http://blog.trueknowledge.com/2009/03/true-knowledge-answering-more-and-more.html

I understood your question to mean:
How many bones (a rigid organ in the skeleton of vertebrates) does a human body (the physical aspect of human being - in contrast to the organism conceived as a person) have?
This conclusion is based on a single fact in the knowledge base:
206 is the count for the meronym holonym pair bone and human body - agree / disagree

It will even show you how it pieced together known facts to answer your question. It's pretty neat, although you have to register to be a beta tester to use it as of right now.

Well, quizbot from trueknowledge already does what
wolfram alpha promise to do in May.

http://quizbot.trueknowledge.com/

True Knowledge have been doing this for over a year. Anyone can add facts to their database, and it will attempt to use those facts to infer answers to questions. Its actually very cool, although doesn't yet support such notions as uncertainty.

I have submitted a project to googles 10^100 compatition that will solve the 4 problems with wikipedia he lists.

The project name is wikixandria and the idea is to make a p2p wiki library.

If the project wins 10^100 we will soon have an academy-friendly alternative to wikipedia.

Also lets remember that is also possible to store knowledge in a knowledge base. Like they do at true knowledge and the let a computer answare our questions.

Also take a loke at my "p2p" news-site crowdnews.eu

http://www.trueknowledge.com/

Its really rough on the side, and isn't quite versatile enough yet, but the idea behind it shows promise... It literally parse your query to figure out what you meant in natural language, translate that in something more usuable by a computer, then use a logical bank of keyword to understand the full meaning of your request, then give you exactly the answer you want.

So you can type "How old is XYZ", and it will tell you something like "I understand your query as asking what is the length of time that has passed since the date of birth of XYZ", then give you the exact answer, along with its primary sources.

Its pretty cool, honestly (right now it doesn't understand enough query variations though. Its getting there).

Not sure if there is bubble but one thing I did learn about during the Cuil-crash-and-burn-at-launch fiasco was a new search technology out of Cambridge that is in beta right now. It is named true knowledge, and uses natural language strings for search, and wiki style user submitted knowledge base in conjunction with a standard search engine. It is pretty neat and promising search technology that I found searching after looking that the shorcomings of Cuil. I highly recommend getting a beta account at true knowledge if you are interested in improving search results in a fine grained approach.

I think trueknowledge is very cool.

http://www.youtube.com/watch?v=IONdWQwcmxA

True Knowledge actually interprets your question using Natural Language Processing, and then looks through a massive database of user-contributed facts, combining them using sophisticated inference rules, to give you the answer you need. Even the inference rules are user-editable.

It's still in beta but http://www.trueknowledge.com/ looks promising, unfortunately I've been waiting to get in on the beta program for a month or so now and all I've gotten is a letter saying they have enough right now but will be in touch in the future.... Haven't really seen any information on an "official" opening date for it either.