I was at Chris Welty’s keynote yesterday at the WWW2012 Conference. His talk was on Jeopardy/Watson and, although this is not the first time I heard/saw something on Watson, some things really became clear only at his keynote. Namely: what is really the central paradigm that made the question answering mechanism so successful in the case of Watson?
Well… query answering in Watson is not some sort of a deterministic algorithm that turns a natural language question into a query into a huge set of data. This approach does not work. Instead, a question is analyzed and, based on search in various set of data, a large set of possible answers is extracted. These “candidate” answers are analyzed separately along a whole series of different dimensions (geographical or temporal dimensions, or, which I found the most interesting, putting back candidate answers into the original question and search that again against various sources of information to rank them again). The result is a vector of numerical values representing the results of the analysis along those different dimensions. That “vector” is summed up into one final value using a weight values for each dimension. The weights themselves are obtained through a prior training process (in this case using a number of stored Jeopardy question/answers). Finally, the answer with the highest value (I presume over a certain threshold value) is returned.
I hope I got it right:-). But the mechanism is certainly something like that. And it is interesting: it is different from the traditional question/answer approaches which is, usually, much more “deterministic”. This is some sort of a new computing paradigm (not necessarily invented by the Watson team, but used by them). Is it a really important new paradigm? Well… to quote Chris: “We won!”.