Woolfram Alpha

I am really interested in seeing this product and how it works, especially in its use of Natural Language. Years ago I was heavily involved in the introduction of what was, at the time, the most advanced, by far, Natural Language processor. The development was carried out in Berkley by two individuals whose conceptual understanding and sheer technical capability was way beyond anyone that I had met before or have met since.

Most so called Natural Language Processors of the time were keyword driven and could only function with a limited set related to those keywords; our product utilised both semantic and syntactic processes (via a very clever parser) and could handle a vast lexicon related to the subject matter contained within one or several databases.

I remember accompanying one of the founders to a meeting with the developers of the leading European project and we found that this product could handle about 10,000 words and really did not have a grammer- ours would have no problem in handling well over 100,000 and had a very powerful grammer.

The difference was that Natural Language (the product name) was not actually interested in words, but in the meaning that a group of words contained. Thus, it was possible to ask the same question in several different ways and get the same answer. It built a context, so from that point it was possible to expand or refine, i.e. broaden the context or refine it to pinpoint a particular thing.

It was (and probably still is but in a different form) an incredible product and one in which I am delighted and proud to have had an involvement. Like all things, however, it had its problems - a few technical and a few operational - but the major issue was getting organisations to understand its power and capability. It is always the same when a revolutionery new product/facility is brought to market: nobody understands and it is rarely the innovator that gains the reward for the innovation but somebody that comes along later.