2. Where Have You Heard of Watson?
• February 14th, 2011 Watson competed on
Jeopardy
• In a three day match, Watson won a one million
dollar prize
• Jeopardy also spent time introducing Watson’s
development and capabilities.
3. What is Watson?
• Watson is a machine that can understand and
answer natural language.
• Basically, Watson acts as a search engine you
can speak to.
• IBM plans to develop Watson beyond Jeopardy,
to advance certain fields.
• Although Watson is very smart, it does not have
the ability to think as a human.
5. What does Watson need for
it to be successful?
• The ability to discern double meanings of words,
puns, rhymes, and inferred hints
• Extremely rapid responses
• The ability to process vast amounts of information
to make complex and subtle logical connections
6. Focus on 3 key capabilities
• Natural language processing
• Hypothesis generation
• Evidence-based learning
7. Development into a
Jeopardy! powerhouse
• 2006: Watson could get 15% of the questions
correct
• 2007: A team of 20-25 people, led by David
Ferrucci, were given 3-5 years to solve the
problems surrounding Watson
• 2008: Watson could compete with Jeopardy!
Champions
• 2010: Watson could regularly beat human
Jeopardy! contestants
12. Hypothesis and Evidence Scoring
• Finds passages from the remaining sources for
positive or negative evidence
• Scoring algorithms rate the quality of these
answers
13. Final Merging and Ranking
• Weigh the importance of the different types of
evidence
• Weighted evidence scores are merged into a final
ranking
▫ If the ranking is below 50%, Watson won’t answer
▫ No fixed confidence level for whether Watson will
answer or not
19. Finance
• Big Data vs. Big Strokes
• Watson vs. Computer Farms
• Patterns vs. Algorithms
Most financial services firms are dying
of thirst in an ocean of data- IBM
Chang
20. Healthcare
• Dr. Watson
• Information database
• Diagnosis and Prescription
"The way we doctors work,…we ask our
colleagues, we ask specialists in the
field. In essence, Watson will have
that capability.” - Dr. Kris Watson
22. Final quote?
Watson is not a deterministic system. Watson
learns, based on the information that's fed to it.
It's also trained; it's a machine-learning
system. It's only as smart as the information it's
fed. It's the way any human being would learn.
24. Based on what we have presented on IBM’s
future expectations of Watson, how do you
feel that this machine will have a large
impact on our futures?
25. What field do you think Watson may have
the largest impact in? Why?
26. What part of Watson’s development do you
find most innovating?
Editor's Notes
These 3 key capabilities that Watson had to learn are inherent to us as human beings. Computers don’t understand words; they understand code written in 0s and 1s. To have Watson listen to a Jeopardy! Question, process the meaning behind the sentence, and actually go through its vast data base to try to find an answer took about 4 years of development by a special team of researchers. Led by David Ferrucci, the senior manager of IBM’s Semantic Analysis and Integration department, Watson improved from 15 % of the questions correct in 2006 to regularly beating human contestants in 2010.
Led by David Ferrucci, the senior manager of IBM’s Semantic Analysis and Integration department, Watson improved from 15 % of the questions correct in 2006 to regularly beating human contestants in 2010.
When the question is given, Watson instantly splits it into parts of speech and tries to identify each of their roles in the sentence as a whole. This helps Watson determine what type of question is being asked and what the question is asking for.
Watson quickly searches through hundreds of millions of documents looking for a few thousand possible answers. At this point in the process, it’s quantity over accuracy because if Watson disregards the correct answer in this process, there’s no way of retrieving it.
Watson finds passages from many different sources to find positive and negative evidence of the remaining possibilities. Watson understands the passages and relationships between those passages. Scoring algorithms rate the quality of these answers on the source material’s reliability to whether the time or location of the answer seem correct. This all happens in a matter of seconds.
Watson uses prior experience to weigh the importance of the different types of evidence. Watson learns how to weigh, apply, and combine its algorithms to help decide the degree each piece of evidence is useful.
IBM Watson has already “wowed” the public with his skills on Jeopardy. As we have been told, Watson has already been hired by Citigroup and by Wellpoint. However, Watson’s career goals reach much farther. The goal is to become a core requirement of these three fields.