$$SWYNK$$
Featured Article(s)
Tips for designing SQL Server 2008 tables (Part 2)
In this article, you can find some useful tips to design SQL Server 2008 tables.
Featured White Paper(s)
Structuring the Unstructured: How to Dimensionalize Semi-Structured Business Data
Written by Interactive Edge
The Business Intelligence industry … (read more)
Implications of IBM Watson
If you heard the presentation of our keynote speaker, Stephen Baker, at the SSWUG DB Tech Conference last spring you may remember him as the author of “Final Jeopardy: Man vs. Machine and the Quest to Know Everything.”
I think this brings into focus the real value of parallel processing and distributed data storage. For those of you who don’t watch USA television, Jeopardy is a game show where contestants are given facts for which they must provide the correct question. For example, an answer might be, “He was the first President of the United States of America.” The contestants must respond with the correct answer of, “Who is George Washington?”
The actual game is nowhere near such a simple set of questions. The questions are broad where esoteric pieces of information provide clues to the requested question. It is a game that demonstrates the amazing nature of the human brain.
IBM developed a computer system called Watson, now old news, capable of winning the Jeopardy show against human participants. This is much bigger than winning a chess game against grand masters. The possibilities are much larger and the evaluation algorithms less simple. In fact, in order to win, the computer must resolve the answer faster than the human competitors. If you have seen the show, you know that responses are most often less than a second.
IBM accomplished this feat with a massively scaled parallel Hadoop engine, using distributed data and NoSQL non structured data stores. I have been talking quite a bit about NoSQL the last few months. However, I never made this connection of the Watson engine to Hadoop and distributed non-structured data storage.
Perhaps that demonstrates further the position of this kind of computing. It isn’t needed for every application. It doesn’t replace structured data storage. There are things it cannot do as efficiently as earlier forms of computing.
It does provide advances in sifting through large volumes of data, stored in a random fashion as simple facts, and the ability to mine that data and make potential connections at an amazing speed. One of the most difficult questions we often try to resolve is Cause and Effect. These are the kind of platforms allowing computers to assist in finding answers to those large questions. Imagine applying it to economics, political science, medicine, ecology, or even insurance.
Watson may not have predicted what new databases will be, but it may well have demonstrated what will be the next step in data mining.
Would you like to play Jeopardy? The Answer is, “The worst thing that could happen to a programmer in the 60s.” Send your answers to btaylor@sswug.org.
Cheers,
Ben