WATSON. Michael Dundek Industry Architect. Best Student Recognition Event July 6-8, 2011 EMEA IBM Innovation Center La Gaude, France

Transcription

1 WATSON Michael Dundek Industry Architect Best Student Recognition Event July 6-8, 2011 EMEA IBM Innovation Center La Gaude, France

2 Want to Play Chess or Just Chat? Chess A finite, mathematically well-defined search space Limited number of moves and states All the symbols are completely grounded in the mathematical rules of the game Human Language Words by themselves have no meaning Only grounded in human cognition Words navigate, align and communicate an infinite space of intended meaning Computers can not ground words to human experiences to derive meaning

3 Taking on Jeopardy! A Grand Challenge On February 14, Watson challenged Jeopardy! world champions Ken Jennings and Brad Rutter in a twomatch contest to be aired over three consecutive nights. Jeopardy! - an American quiz show - covers a broad range of topics, such as history, literature, politics, arts and entertainment, and science. Jeopardy! poses a grand challenge for a computing system: Broad range of subject matter Speed of accurate responses and confidence Requires analyzing subtle meaning, irony, riddles, and other complexities. Beyond Jeopardy!, the technology behind Watson can be adapted to solve business and societal problems for example, diagnosing disease, handling online technical support questions, and parsing vast tracts of legal documents - and to drive progress across many industries.

4 Informed Decision Making: Search vs. Expert Q&A Decision Maker Has Question Distills to 2-3 Keywords Reads Documents, Finds Answers Decision Maker Finds & Analyzes Evidence Asks NL Question Considers Answer & Evidence Search Engine Finds Documents containing Keywords Delivers Documents Expert based on Popularity Understands Question Produces Possible Answers & Evidence Analyzes Evidence, Computes Confidence Delivers Response, Evidence & Confidence 4

5 Hard Questions? Computer programs are natively explicit, fast and exacting in their calculation over numbers and symbols.but Natural Language is implicit, highly contextual, ambiguous and often imprecise. Where was X born? Structured Unstructured One day, from among his city views of Ulm, Otto chose a water color to send to Albert Einstein as a remembrance of Einstein s birthplace. X ran this? If leadership is an art then surely Jack Welch has proved himself a master painter during his tenure at GE.

6 What is Watson? The first non-human contestant on the quiz show Jeopardy!, Watson is a workload optimized computer system that battled the game s best human players A computing system built by a team of IBM scientists who set out to create a system that rivals a human s ability to answer questions posed in natural language with speed, accuracy and confidence A system with the capability to understand the meaning and context of human language, and rapidly process information to find precise answers to complex questions which holds enormous potential for businesses Of course a lot more is at stake than just a game show victory Researchers have their sights set on applying the technology in fields from health care to help desks It's easy to see how such a breakthrough could be put to good use there, helping doctors to accurately diagnose patients' conditions by sifting through mountains of data in mere seconds. USA Today, 1/14/11

7 The Jeopardy! Challenge: A compelling and notable way to drive and measure the technology of automatic Question Answering along 5 Key Dimensions Broad/Open Domain Complex Language $200 If you're standing, it's the direction you should look to check out the wainscoting. $1000 The first person mentioned by name in The Man in the Iron Mask is this hero of a previous book by the same author. High Precision Accurate Confidence High Speed $600 In cell division, mitosis splits the nucleus & cytokinesis splits this liquid cushioning the nucleus $2000 Of the 4 countries in the world that the U.S. does not have diplomatic relations with, the one that s farthest north

8 Broad Domain We do NOT attempt to anticipate all questions and build databases. We do NOT try to build a formal model of the world In a random sample of 20,000 questions we found 2,500 distinct types*. The most frequent occurring <3% of the time. The distribution has a very long tail. And for each these types 1000 s of different things may be asked. Even going for the head of the tail will barely make a dent 8 *13% are non-distinct (e.g, it, this, these or NA) Our Focus is on reusable NLP technology for analyzing vast volumes of as-is text. Structured sources (DBs and KBs) provide background knowledge for interpreting the text.

9 Different Types of Evidence: Keyword Evidence In May 1898 Portugal celebrated the 400th anniversary of this explorer s arrival in India. In May, Gary arrived in India after he celebrated his anniversary in Portugal. celebrate d Keyword Matching arrived in celebrated In May 1898 Keyword Matching In May 400th anniversary Keyword Matching anniversary Evidence suggests Gary is the answer BUT the system must learn that keyword matching may be weak relative to other types of evidence 9 arrival in India explorer Portugal Keyword Matching Keyword Matching Gary India in Portugal

10 Different Types of Evidence: Deeper Evidence In May 1898 Portugal celebrated the 400th anniversary of this explorer s arrival in India. May 1898 Stronger evidence can be much harder to find and score. celebrate d 400th anniversary arrival in India Portug al Search Far and Wide Explore many hypotheses Find Judge Evidence Many inference algorithms Temporal Reasoning Statistical Paraphrasi ng GeoSpatial Reasoning On On 27th 27th May May 1498, 1498, Vasco Vasco da da On Gama the 27th 27 Gama landed landed May th of 1498, May in in Kappad Kappad Vasco 1498, da Beach Beach Vasco Gama landed da Gama in Kappad landed Beach in Kappad Beach Date Math Paraphrase s Geo- KB landed in Kappad Beach 27th May explorer The evidence is still not 100% certain. Vasco da Gama

11 DeepQA: The Technology Behind Watson Massively Parallel Probabilistic Evidence-Based Architecture DeepQA generates and scores many hypotheses using an extensible collection of Natural Language Processing, Machine Learning and Reasoning Algorithms. These gather and weigh evidence over both unstructured and structured content to determine the answer with the best confidence. Learned Models help combine and weigh the Evidence Question Answer Source s Primary Search Candidate Answer Generation Answer Scoring Evidence Sources Evidence Retrieval Deep Evidence Scoring Model s Model s Model s Model s Model s Model s Question & Topic Analysis Question Decomposition Hypothesi s Generatio n Hypothesis and Evidence Scoring Synthesis Final Confidence Merging & Ranking Answer & Confidence

12 Confidence is King All questions are not equal some take longer, some are less certain than others Watson ring-ins ONLY IF it can compute a confidence fast enough It considers the category, question and a self-assessment of its own algorithms Uses confidence to make betting decisions and mange the risk associated with possibly getting wrong answers 12

13 DeepQA: Incremental Progress in Answering Precision on the Jeopardy Challenge: 6/ /2010 IBM Watson Playing in the Winners Cloud v0.8 11/10 V0.7 04/10 v0.6 10/09 v0.5 05/09 v0.4 12/08 v0.3 08/08 v0.2 05/08 v0.1 12/07 Baseline 12/06

14 Precision, Confidence & Speed Deep Analytics Combining many analytics in a novel architecture, we achieved very high levels of Precision and Confidence over a huge variety of as-is content. Speed By optimizing Watson s computation for Jeopardy! on over 2,800 POWER7 processing cores we went from 2 hours per question on a single CPU to an average of just 3 seconds. Results in 55 real-time sparring games against former Tournament of Champion Players last year, Watson put on a very competitive performance in all games -- placing 1 st in 71% of the them!

15 Future of Watson Apply it to Business Applications Scale it Make it smaller Open it up to third parties

16 Watson It s about much more than a quiz show: What it means for business A new paradigm in IT The computing paradigm for business has changed traditional computing models are being replaced by systems that underlie every business process, and the hardware and software performance in those systems is closely tied to actual business performance Watson harnesses IBM s commercially available, workload optimized POWER7 system which can process thousands of simultaneous tasks at rapid speeds ideal for complex analytics workloads. The future of business is in the data Watson-like analytics applied to business can provide answers with a confidence ranking that can be gleaned from both structured and unstructured data by running hundreds of different kinds of analytical queries across all different kinds of information Applying those innovations from Watson to an organization can help transform business models meaningful insights from information can help anticipate and shape better business outcomes, improve business operations and boost service to customers A smarter way to get things done Beyond Jeopardy!, the technology behind Watson can be adapted to help solve business and societal problems diagnose disease, handle online technical support, parse vast tracts of legal documents Watson-like capabilities applied to healthcare, government, transportation, and other industries can enable Smarter Planet transformations

17 From battling humans at Jeopardy! to transforming business To compete at Jeopardy!, humans and computers need to: Tap into a broad, open domain of clues Parse complex human language Be highly precise at finding the answer and put a confidence behind that answer Watson-like capabilities can now be applied to your business: Manage massive amounts of data from business operations and an instrumented world Deal with structured and unstructured data to gain meaningful insight into your business Have a high level of confidence in analytics to make key business decisions Do all of that in under three seconds Do it at the speed of business, in real time

18 Potential Business Applications Telco s Host Healthcare / Life Sciences: Diagnostic Assistance, Evidenced-Based, Collaborative Medicine for their customers and Healthcare Professionals Telco Tech Support : Help-desk, Contact Centers, Self Service, Cellular Plan Selection for Customer and CSR, Information services Support Repair Staff: Use to support trouble shooting of network and customer premise faults Telco s Host Government Services: Improved Information Sharing 18

19 DeepQA in Continuous Evidence-Based Diagnostic Analysis Symptoms Diagnosis Models Renal failure Find Meds Hist Fam Symp Confidence Family History Patient History Medications Tests/Findings UTI Diabetes Influenza Notes/Hypotheses hypokalemia esophogitis Most Confident Diagnosis: Influenza Most Confident Diagnosis: Diabetes Most Confident Diagnosis: UTI Huge Volumes of Texts, Journals, References, DBs etc.

20 Watson Fun Facts Watson digested 1 million books about 200 million pages of information Wikipedia, Bible, Internet media data base, NY times, Shakespeare, World Book encyclopedia Watson consumes 85 thousand watts ( Human brain consumes 20) Watson is made up of 10 rack of Power 7 computer holding 90 servers Watson is run by about 2800 preocessors It has a footprint of 800 cubic feet qnd weighs 9 tons ( Human brain weighs 3 pounds) It requires 40 tons of cooling It has 15 TB of RAM

21 21

22