Wolfram Alpha is a knowledge computation engine from Wolfram Research which enables one to find answers to his question in the form of a concise factual page often with interesting data visualization methods. It’s a result of work of over two years by a team of about 100 people who manually curated most of the data and worked on many linguistic problems which enables engine to understand all the possible ways in which user may query the data. It uses Mathematica at it’s core to make sense of the information fed to it.
It’s a service (not so much a web search engine) and not the only one among alternative search engines who all believe in one thing Search is much more than searching just hyper linked pages. It has to be seen as data at it’s lowest level which has this property that the way they are linked to other web pages largely defines it’s importance and many other dimensions. There have been Realtime search engines,Semantic Search Engines, Social Search engines and a lot more.
So ,coming back to wolfram alpha. It has been a daunting task in itself ,and got the hype it deserved. Though the question remains will it be able to go mainstream such that a average user can understand it ,or will it become just another cool service which no one really uses. Though sometimes the engine not sure about the input, other times the results are not relevant . Given the alpha status and what it is trying to achieve it can be ignored for now.
But for sure this is only the start of things to come. So far the data they have been using have been curated manually and made sure that the data is correct .The first big task will be to automate this process which seems to be a very big a problem in itself. How do i decide while one fact is right , while other is wrong. It’s does not even to be wrong all the time , just less relevant .Restricting data at the source level is one way that can be done .But is it the right way? I don’t know.
Another Problem is that is you get the answer(s) if you ask a question (hmmm the right question 😉 ), and if engine is somehow automagically able to choose the right sense ( it’s a word , related to finance ) . There are times when i don’t know what exactly i am looking for, hell there are times when i don’t even know the right word, google suggest comes to rescue. In this case google gives me the option of choose among a number of choices to build my query starting with not even having the right word. You may not realise it but that’s what the one does end up doing up most of the times. It comes from having way more data and then building clusters on top on it about the search word. I also get a good help from partial search results . This exploration part atleast for now does not seem to be possible until one gets the first right hit.
So it has just started and the possibilities are limitless , and i expect that there will be lots of new interesting ways it will be used .It can be something we don’t know about right now as you can define it as you want it.