<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"><head><meta http-equiv=Content-Type content="text/html; charset=us-ascii"><meta name=Generator content="Microsoft Word 12 (filtered medium)"><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Consolas;
panose-1:2 11 6 9 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:11.0pt;
font-family:"Calibri","sans-serif";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
p.MsoPlainText, li.MsoPlainText, div.MsoPlainText
{mso-style-priority:99;
mso-style-link:"Plain Text Char";
margin:0in;
margin-bottom:.0001pt;
font-size:10.5pt;
font-family:Consolas;}
span.EmailStyle17
{mso-style-type:personal-compose;
font-family:"Calibri","sans-serif";
color:windowtext;}
span.PlainTextChar
{mso-style-name:"Plain Text Char";
mso-style-priority:99;
mso-style-link:"Plain Text";
font-family:Consolas;}
.MsoChpDefault
{mso-style-type:export-only;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style><!--[if gte mso 9]><xml>
<o:shapedefaults v:ext="edit" spidmax="1026" />
</xml><![endif]--><!--[if gte mso 9]><xml>
<o:shapelayout v:ext="edit">
<o:idmap v:ext="edit" data="1" />
</o:shapelayout></xml><![endif]--></head><body lang=EN-US link=blue vlink=purple><div class=WordSection1><p class=MsoPlainText>Kelly, thanks. These are excellent questions, which I'll include in a follow-up interview. We just posted three IBM videos that discuss customer service, finance, and healthcare applications; and two more on other Watson design issues, including one related to building the system: <o:p></o:p></p><p class=MsoPlainText><a href="http://www.kurzweilai.net/videos">http://www.kurzweilai.net/videos</a>.<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>-----Original Message-----<br>From: Kelly Anderson [mailto:kellycoinguy@gmail.com] <br>Sent: Sunday, February 13, 2011 9:57 PM<br>To: amara@kurzweilai.net; ExI chat list<br>Subject: Re: [ExI] Watson on NOVA<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>On Sun, Feb 13, 2011 at 2:14 AM, Amara D. Angelica <amara@kurzweilai.net> wrote:<o:p></o:p></p><p class=MsoPlainText>> Kelly, I had similar questions, so I interviewed an IBM Watson research<o:p></o:p></p><p class=MsoPlainText>> manager. Please see if this helps:<o:p></o:p></p><p class=MsoPlainText>> http://www.kurzweilai.net/how-watson-works-a-conversation-with-eric-brown-ib<o:p></o:p></p><p class=MsoPlainText>> m-research-manager. I would be interested in any critiques of this, or<o:p></o:p></p><p class=MsoPlainText>> questions for a follow-up interview.<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>"open, pluggable architecture of analytics" sounds like it has an<o:p></o:p></p><p class=MsoPlainText>engine, and can add heuristics. If that's the case, then this is a<o:p></o:p></p><p class=MsoPlainText>pretty powerful core technology, but it requires that it be "built and<o:p></o:p></p><p class=MsoPlainText>tuned to play Jeopardy!" So if I were going to ask follow up<o:p></o:p></p><p class=MsoPlainText>questions, I would ask some along these lines...<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>On the NOVA show it talked about adding gender information... is this<o:p></o:p></p><p class=MsoPlainText>one of the pluggable pieces you are referring to?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>When you say "open" do you mean open source? Or open for purchasers of<o:p></o:p></p><p class=MsoPlainText>the system to augment?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>Is this going to be available in a cloud configuration anytime soon?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>Tell us more about "building" and "tuning"... It appears from the NOVA<o:p></o:p></p><p class=MsoPlainText>show that it took 4 years to build and tune the system for Jeopardy,<o:p></o:p></p><p class=MsoPlainText>how much effort would it take to build and tune a system for medical<o:p></o:p></p><p class=MsoPlainText>diagnosis? Or build a technical support database for say Microsoft<o:p></o:p></p><p class=MsoPlainText>Word.<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>It seems that the natural language processing of the questions and<o:p></o:p></p><p class=MsoPlainText>categories is very extensive and uses a kind of search tree technology<o:p></o:p></p><p class=MsoPlainText>reminiscent of AI search trees used in games such as chess. Is that<o:p></o:p></p><p class=MsoPlainText>correct?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>Tell us more about the index that is build a priori of the raw data<o:p></o:p></p><p class=MsoPlainText>that the answers are sought from. Is it indexed, or is there just a<o:p></o:p></p><p class=MsoPlainText>brute force algorithm based on keyword searches and then further<o:p></o:p></p><p class=MsoPlainText>statistical processing of the results of the keyword search. In other<o:p></o:p></p><p class=MsoPlainText>words, what's done prior to the question being asked on the index side<o:p></o:p></p><p class=MsoPlainText>of the equation?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>(I'm sure you could make that question shorter... :-)<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>You talk about Watson "learning", is the learning on the side of<o:p></o:p></p><p class=MsoPlainText>understanding the question, finding the answer or both? Are you using<o:p></o:p></p><p class=MsoPlainText>neural networks, statistical approaches, or some new approach for<o:p></o:p></p><p class=MsoPlainText>that?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>If developers wanted to build and tune their own solutions on this<o:p></o:p></p><p class=MsoPlainText>architecture, how soon do you think it will be available? Is there a<o:p></o:p></p><p class=MsoPlainText>business unit working on this yet?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>Are there going to be any papers published by the Watson team?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>What aspect of Watson is the most novel? Or is Watson just putting<o:p></o:p></p><p class=MsoPlainText>together the best of what was already out there in a really good way?<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>I'm sure I could come up with more questions... but those would be<o:p></o:p></p><p class=MsoPlainText>among the ones I would ask first I think. I really liked your article.<o:p></o:p></p><p class=MsoPlainText>It was particularly interesting to listen to them think about what<o:p></o:p></p><p class=MsoPlainText>IBM's business model for such things might be.<o:p></o:p></p><p class=MsoPlainText><o:p> </o:p></p><p class=MsoPlainText>-Kelly<o:p></o:p></p><p class=MsoNormal><o:p> </o:p></p></div></body></html>