LBi London

Wolfram Alpha introduction webinar

Posted by Manley May 5th, 2009

Stephen Wolfram is getting very close to the launch of the Wolfram Alpha ‘knowledge engine’. Today he is chairing an interactive webinar of his engine so far.

So, if Wolfram Alpha is to be a Google killer, just how does the knowledge engine compare with Google itself?

As a knowledge and data manipulation device, Wolfram Alpha looks to be superb. Frankly I can imagine myself using this every day. But as a search engine replacement?

There has always been talk of users not visiting websites in the future, but that everyone will have a personal "agent" which will go out and collect and manipulate all the data about any presented question, formatting it in a manner that the user can read.

This will range across the board, from buying a ticket to an Exeter City game to a full blown knowledge search.

Wolfram Alpha seems to be a good step towards that paradigm. As has been joked already, it is Tim Berners Lee’s pedantic web all over again, but perhaps a practical one? From initial glances, Wolfram Alpha certainly looks promising.

In the mean time, I am going to listen to the rest of this webinar – so far it has followed my previous experiences of the engine quite closely, but I look forward to the Q&A.

Currently only to be presented in English – future plans to automate identifying of languages, but for now enough of a challenge to simply manage English.

Q. NASA has been using ”NGC1333 IRAS 4a” to test various systems – be curious how it would do here

I was most impressed how the engine compared data between this satellite and nebular.

Q. How does Wolfram|Alpha deal with biases in authorship? For example, how would one ensure neutrality on a page like ”Israel”?

This seems to be a Curation issue, Stephen insists that a level of human intervention will always be required. I find the avoidance on bias to be a major issue for WA, just as it is for Wikipedia.

Q. Will Alpha be freely available or a subscription service?

Both, all that has been shown to us will be free, but there will also be a pro model.

Q. Will there be an API?

Yes, information about the API is already available, but there will be several levels of API available.

Q. What is the anticipated business model?

Corporate sponsors, corporate partnerships, targeted advertising, vendor specific information based on queries and subscription based APIs and professional version, including using WA technology to build bespoke engines. For web search and companies with search and news based models a partnership model looks very good. Currently WA is a hot topic within the industry and some obvious companies.

This did sound a lot like a purchase model, so perhaps not so much a Google killer as a Google acquisition?

Q. Could a company or manufacturer contribute data about their products?

Essentially the answer was yes.

Q. If there is an emphasis on client side processing, how does Wolfram Alpha cope with accessibility standards and the DDA?

No answer.

Tags: wolfram alpha

0 comments Share

A British Google killer?

Posted by Rik Stewart March 11th, 2009

It is a fairly regular and predictable event now in the search industry. A new search engine product surfaces, the community gets worked up into a frenzy, we all hail the new Google killer and then it all goes away. Can a British offering fare any better?

Anyone remember Wikia, Cuil or Squidoo? While Cuil is by no means entirely out of the running, it has not gained the traction that some would have liked to see.

Now there is news of a new search engine from Britain all set to take its place in the semantic web. As reported at Guardian.co.uk a search engine called Wolfram Alpha is due for launch in May 2009. What is different about Wolfram is that it claims to use Natural Language Processing (NLP).

The majority of search engines use keyword search and match the search query to pages contained in a database, albeit using fairly sophisticated algorithms and filters. An NLP based search engine would allow the user to type in questions and receive answers in the form of a list of web pages that hopefully have the information the user is looking for.

NLP-based search engines are not new either, there have been more than a few. True Knowledge based in Cambridge UK, was launched into private beta in November 2007, Powerset was unveiled in September 2007 and Lexxe was founded in 2005, to name but a few.

So, can Wolfram be a Google killer? Unlikely in my opinion. The first hurdle is getting noticed. There are thousands (and that is a very conservative estimate) of search engines out there, many (if not most) of which are little fish all nipping away trying to get a bit of the market. Secondly, to even get close to Google, a company would have to have a budget big enough to take it on. Yahoo!, MSN and Ask all have spent a considerable amount on this very problem and have not really got very far.

However, as a business move, creating such a search engine could be good. After all, if the product is good enough it will attract attention. Powerset is a good example of this. On July 1, Microsoft signed an agreement to acquire it for a reported $100M.

I do not suppose for a moment that Stephen Wolfram is building a computational data engine with a new approach to knowledge extraction and an easy-to-use interface simply to sell it but, if his algorithm is really as good as is claimed, incorporating it into other engines can only be good for the user.

Tags: Google, wolfram alpha

1 comments Share