Most of us recall the quote, “Elementary, my dear Watson,” which is attributed to the Sherlock Holmes character replying to his friend and associate, Dr. Watson.
It is surprising to most people when they learn the Holmes character never actually spoke this line to Dr. Watson. Holmes did say the words, “Elementary” and “My dear Watson,” but never in one complete sentence.
Spoken March 10, 1876 by Alexander Graham Bell to his assistant Thomas A. Watson, the following quote has special meaning for yours truly. “Mr. Watson! Come here! I want to see you!” In 1931, Watson is reported to have said he recalled the words spoken by Bell as, “Mr. Watson, come here, I want you.”
By now you may be wondering what I am leading up to with all this Watson talk.
It’s elementary, my dear readers.
The giant computer maker, IBM, has come up with a new super computer, which recently competed in a practice round against the two top-ranked players of the popular television game show “Jeopardy!”
This new computer is called “Watson” and is named after IBM founder Thomas J. Watson.
Speaking in a relaxed manner during the practice round, the sentences Watson used were fluid, not choppy or “computer-like” at all.
Watson was able to instantly respond (in the form of a question) to the answers as displayed on the “Jeopardy!” board.
The looks of frustration displayed at times on the faces of the human players standing on either side of Watson had me wondering if this computer should have been instead called “HAL” as in reference to the sci-fi movie “2001: A Space Odyssey.” This futuristic computer, HAL 9000, combatingly challenges and almost outsmarts his human counter-parts.
Instead of HAL, IBM says Watson is more like the computer on Star Trek.
Watson’s hardware is made up of 10 racks of IBM POWER 750 servers using the Linux operating system inside 10 refrigerator-sized cabinet bays.
Watson has 15 terabytes of Random Access Memory (RAM), 2,880 processor cores, and operates over a massively-parallel high-performance computing platform.
Watson also has the ability to learn.
The Open Advancement of Question Answering (OAQA) is an area of computer science which develops software models that are able to provide precise and informative answers to questions asked of it using language, such as English, spoken in a natural manner.
The IBM Research computer science paper titled “Towards the Open Advancement of Question Answering Systems” discussed this software model.
Dr. Bill Murdock is a computer scientist with IBM who is involved with the technology’s algorithms or set of rules used in Watson, co-wrote this paper with other researchers in February of 2008.
The paper goes into details about the challenges which needed to be overcome for successful development of an OAQA algorithm to be used in the software code.
This is not just a typical software search and display program, but an advanced and revolutionary means of having a computer actually “understand” the question being asked, thus providing a more complete and informative answer.
The answers could be acquired from resources found in scattered or semi-structured Internet web pages and blog postings or from more structured and precise data found in online encyclopedias, and educational, governmental, and similar online or private databases.
One section of the paper talks about developing a Challenge Set Profile for testing the software’s Question Answering solution abilities.
One of the five challenge problems was identified as the “Jeopardy!” game show.
This IBM paper states that in order for the computer program to succeed in this challenge problem, the software coding itself would require high degrees of accuracy, speed, and confidence.
Yes, software having “confidence” is a new one for me, too.
The paper describes how sophistication would be needed in working out the software program’s grammatical structure and phrasing of sentences used in answers to the sometimes wordy “Jeopardy!” questions. The IBM paper acknowledges this “natural language parsing” between humans and the computer would be a difficult challenge.
Scientists at IBM worked for four years on this highly- advanced Question Answering (QA) system for Watson, calling it DeepQA.
Watson competed against the two top-ranked “Jeopardy!” contestants in a practice round this past January.
This practice round shows Watson being represented physically by a display screen with its avatar flickering and flashing in a manner IBM described as “a global map projection with a halo of ‘thought rays.’” To the left and right of Watson stood Jeopardy! champions Ken Jennings and Brad Rutter.
Watson acted very human, as it patiently listened to each question and then buzzed-in with a verbal response.
For each question, Watson simultaneously processed thousands of its algorithmic formulas via its sophisticated DeepQA program.
Watson performs 80 trillion calculations or operations per second.
During this practice round, Watson was not directly connected to the Internet, but was instead accessing millions of pages of previously scanned content which included books, entire encyclopedias, databases of literature, and other sources.
Watson ended up defeating its human opponents during this practice round.
This reminded me of when IBM’s Deep Blue computer defeated chess grandmaster Garry Kasparov back in 1997.
Watson participates in an official “Jeopardy!” tournament-style competition airing Monday, Feb. 14 through Wednesday, Feb. 16.
Watch Watson’s Jeopardy! practice round at http://tinyurl.com/6bsusvq.
Also, check out an in-depth interview with a member of IBM’s Algorithms Team at http://tinyurl.com/4rrsv3g.