Friday, February 15, 2008

True Knowledge - Computer System that Understands

Although Valentine’s Day has ended, it would be interesting if I can start this blog with Valentine-related topic. I started to find a topic by doing a search with “love” + “computer” keywords on Google. Indeed, I found something interesting, which is a blog posting titled “Teaching a Machine to Understand love”. This blog discussed about a blogger who tried to teach his computer system what love is. The first question popped up on my mind is “How to do that?”. This Question leaded me to a system with unique technology called “True Knowledge”.

According to the developer, True Knowledge is “a technology which can represent the world's knowledge in a form that is clear and accessible to humans, as well as being comprehensible to computers”. That means that this technology enables the knowledge to be stored in the form that can be understood by computer, not merely plain text that it doesn't understand. Thus, computers can analyse and process that knowledge and then derive new knowledge.

This knowledge can then be used to answer questions from users or other computers. The developer claimed that this system “can answer questions it hasn't seen before and can combine knowledge through a process of inference and cross-referencing stored information to produce a reasoned answer.” This makes the system different from standard search engines such as goggle or yahoo. Standard search engines response to queries or questions by searching for web pages that have matching keywords, and then it returns the list of the matched web pages as search result. Therefore, most of the time users may get irrelevant responses.

Unlike the standard search engine, this system intelligently addresses users’ queries or questions by providing straight-forward answer. At the same time, it also provides standard search result as supplementary response. The following picture is a picture that I took from the developer website. It shows a sample response of the question: “Is Jennifer Lopez Single?”




Knowledge gets in to the system from two main sources: users' contribution and external databases. Users can contribute new knowledge to the system, just like Wikipedia. To contribute, users only required to type in facts in plain English. Users can also modify or remove incorrectly or maliciously added fact. Both knowledge from users and external database are analysed and processed before they are stored to the knowledge base.

On the main page of the developer website, there is a form in which visitor can sign up to test this technology. So I sign up, hoping to get a chance to try the system myself. However, I felt so disappointed after I had completed the registration process. I got a notification saying: “We'll be in touch with an invitation to create an account in due course, however please note that there are currently many thousands of people in the queue, so there may be a delay while we scale up the test.”

Although I could not test the system myself, I found a demo video on You Tube. The video shows sufficient information about the technology, including: the features, architecture, comparisons with other technology and etc. Here I included the video.



This technology may replace the standard search engine in future time. I suppose, if "True Knowledge" managed to launch this service, it will be a threat to Google and other standard search engine companies. And this makes me wonder: “How come Google, as a current biggest search engine company, has not shown any sign of movement to compete?” or "are they silently preparing their 'secrete weapon'?".


References

4 comments:

PBarnaghi said...

Well done!
CyC is a (similar?) projet which tries to create a comprehensive ontology. Plese see:
http://en.wikipedia.org/wiki/Cyc

Payam

Zalina said...

wow! i never know about this, seems to be interesting!

Anonymous said...

I think google has done something undercover. Quite impossible if they just wait to be beaten.

Edy said...

Yes Dr. Payam, I think they are similar. But not necessarily the same. this following quote shows that it has similar concept with True Knowledge.
"Typical pieces of knowledge represented in the database are 'Every tree is a plant' and 'Plants die eventually'. When asked whether trees die, the inference engine can draw the obvious conclusion and answer the question correctly."
However, I think they might be different in the way they are implemented.