On Mon, February 14, 2011 19:41, Patrick Cassidy said:
> Doug Foxvog wrote: (01)
>> I'll be interested in what questions it does not answer. I don't think
>> we'll have any way of knowing if it "clicks" late. (02)
> The display tonight showed Watson's top 3 answers, with confidence, and in
> at least one case it had the right answer, at above the minimum confidence
> level, but didn't get the click. So it apparently clicked late in that
> case. (03)
Was that display up by the time the other contestant clicked in? (04)
> I liked the case where Watson answered "finis" for a Latin-derived word
> meaning ending where a train can also start. That was about the only one
> whose answer I knew - "terminal". Also, Watson gave "the 1920's" when the
> answer ("the twenties") the decade) had already been declared wrong. (05)
It had been explained that Watson's only input was textual. Watson was
not receiving oral input. The confidence level it had for "the 1920s"
did not change when the incorrect answer, "the twenties" was given. (06)
Watson's second best guess for the answer to that question was "1923" --
which is not a decade. This further demonstrates that Watson is not
attaching a meaning to its answers. (07)
In the Harry Potter question, who was the killer of various victims,
Watson ranked "Harry Potter" higher than "Voldemort". This is probably
because the word "Harry Potter" is more closely associated with the
others *and* because in the books, Voldemort's name is avoided by the
characters. (08)
All three segments were probably taped in a single day. There would
have been no option for programming between the various segments. (09)
-- doug f (010)
> I
> wonder what lessons Watson's developers will derived from those blunders.
>
> In any case, very impressive!!! Congratulations to the whole Watson team.
> Looking forward to more.
>
> Pat
>
> Patrick Cassidy
> MICRA, Inc.
> 908-561-3416
> cell: 908-565-4053
> cassidy@xxxxxxxxx
>
>> -----Original Message-----
>> From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-
>> bounces@xxxxxxxxxxxxxxxx] On Behalf Of doug foxvog
>> Sent: Monday, February 14, 2011 12:46 PM
>> To: [ontolog-forum]
>> Subject: Re: [ontolog-forum] IBM Watson on Jeopardy
>>
>> On Mon, February 14, 2011 12:21, Ron Wheeler said:
>> > So we are now worried that a computer is going to be better than
>> humans
>> > at forecasting how a question will be finished?
>>
>> > I will be completely impressed (blown away actually) if Watson has
>> the
>> > answer before the question is asked and is bouncing up and down
>> > screaming "I know what you want to know and I already have the
>> answer".
>>
>> The question (actually in Jeopardy terms the "answer") is presented in
>> text form, which is then read allowed. Not only do humans often have
>> the
>> answer long before the reading is finished, Watson also often has had
>> its answer finished (in trial runs) well before the reading is complete.
>> Both the humans and Watson must wait until the right time before
>> signaling that they have an answer.
>>
>> > Much higher expectations than I originally had!
>>
>> I'll be interested in what questions it does not answer. I don't think
>> we'll have any way of knowing if it "clicks" late.
>>
>> > Now we have the ultimate computer. It knows the questions and the
>> > answers - we just have to sit back and watch the world roll on.
>>
>> Hah!
>>
>> > Ron
>> >
>> > On 14/02/2011 11:42 AM, doug foxvog wrote:
>> >> One key advantage that Watson has is timing. One can not push the
>> >> button
>> >> to signal a question as soon as one guesses it, but must wait until
>> the
>> >> spoken answer is finished. If one pushes the button too soon, there
>> is
>> >> a
>> >> minimum delay before one can push the button again. A contestant
>> can
>> >> not
>> >> judge the timing to the microsecond, but Watson probably can. This
>> >> seems
>> >> to be a great advantage for the computer. Perhaps Watson should
>> signal
>> >> when it has found a question (well before the spoken answer has
>> finished
>> >> being read), and then resignal at the delay time to which a human
>> >> contestant is limited.
>> >>
>> >> One often sees that several contestants on Jeopardy are pushing the
>> >> button,
>> >> but one wins on timing. How's that going to work out on Jeopardy?
>> >>
>> >> -- doug f
>> >>
>> >> On Sun, February 13, 2011 17:25, Patrick Durusau said:
>> >>> Jack,
>> >>>
>> >>> On 2/13/2011 5:00 PM, Jack Park wrote:
>> >>>> Returning to the original thread, I've noticed that the Web is
>> crowing
>> >>>> over the fact that Watson is running under Linux.
>> >>>>
>> >>> Do you know if Linux was chosen so the Watson team would not have
>> to
>> >>> pay
>> >>> licensing fees to the AS/400 division? ;-)
>> >>>
>> >>> Minsky did not sound bitter to me.
>> >>>
>> >>> Disappointed that the news media has bestowed the AI crown on a
>> system
>> >>> with no more understanding of Jeopardy than an orrery has of
>> >>> astrophysics, but not bitter.
>> >>>
>> >>> Disappointment is understandable when a lifetime of work is made
>> into a
>> >>> carnival sideshow type event. Jeopardy no more measures
>> intelligence
>> >>> than a standard IQ test. Both measure being good at the skills
>> required
>> >>> by each one. Nothing more, nothing less.
>> >>>
>> >>> Hope you are having a great weekend!
>> >>>
>> >>> Patrick
>> >>>> Jack
>> >>>>
>> >>>> On Sun, Feb 13, 2011 at 9:35 AM, Peter Yim<peter.yim@xxxxxxxx>
>> >>>> wrote:
>> >>>>> Thank you, Bill.
>> >>>>>
>> >>>>> I concur with Bill Andersen that this spun-off conversation about
>> the
>> >>>>> military or politicians should be taken elsewhere. So, please.
>> >>>>>
>> >>>>> The subject is "iBM Watson on Jeopardy" and that's a fascinating
>> >>>>> topic
>> >>>>> ... continue on that by all means.
>> >>>>>
>> >>>>> Thanks& regards. =ppy
>> >>>>>
>> >>>>> Peter Yim
>> >>>>> Co-convener, ONTOLOG
>> >>>>> --
>> >>>>>
>> >>>>>
>> >>>>> On Feb 13, 2011 7:56 AM, "Bill Andersen"<andersen@xxxxxxxxxxxxx>
>> >>>>> wrote:
>> >>>>>> Ahem... Anyone in this conversation been in the military or have
>> >>>>>> credentials in military history? One good reason I'd rather not
>> see
>> >>>>>> this conversation on ontolog. Another is that it's way out of
>> scope
>> >>>>>> and bordering on political - and those of you who know me know
>> you
>> >>>>>> don't want me to go there.
>> >>>>>>
>> >>>>>> So please take this somewhere else (but include me - I'd love to
>> get
>> >>>>>> into this one)
>> >>>>>>
>> >>>>>> Bill
>> >>>>>> On Feb 13, 2011, at 8:18, "John F. Sowa"<sowa@xxxxxxxxxxx>
>> wrote:
>> >>>>>>
>> >>>>>>> Vincent and Rich,
>> >>>>>>>
>> >>>>>>> JFS
>> >>>>>>>>> That is also true of people. As the saying goes, generals are
>> >>>>>>>>> very good at fighting the last war, but not the current one.
>> >>>>>>> VW
>> >>>>>>>> If that saying were true, there'd be ZERO
>> >>>>>>>> national security - for any country.
>> >>>>>>> You need to study military history. The leaders who win
>> >>>>>>> are the ones who can innovate to meet changing conditions.
>> >>>>>>> But most leaders in every field aren't innovative -- they
>> >>>>>>> just repeat what worked last time.
>> >>>>>>>
>> >>>>>>> RC
>> >>>>>>>> we are buying a bunch of F-35's to fight international
>> terrorism.
>> >>>>>>> That is comparable to the French building the Maginot line
>> after WW
>> >>>>>>> I.
>> >>>>>>> It's a hugely expensive waste of money on systems that would
>> have
>> >>>>>>> been useful in the past.
>> >>>>>>>
>> >>>>>>> VW
>> >>>>>>>> I'll just repeat one of my earlier statements: "There are lots
>> >>>>>>>> of sayings among us which should be relegated to the past -
>> >>>>>>>> and left there - where they belong..."
>> >>>>>>> You can repeat that, but it won't make it true.
>> >>>>>>>
>> >>>>>>> A better guideline is one of my favorite quotations from
>> >>>>>>> Alfred North Whitehead:
>> >>>>>>>
>> >>>>>>> "Every great truth is only half true."
>> >>>>>>>
>> >>>>>>> That means that every proverb, principle, or ideology has
>> >>>>>>> to be re-evaluated in terms of current conditions in order
>> >>>>>>> to determine which half is appropriate.
>> >>>>>>>
>> >>>>>>> The most dangerous offenders are politicians who repeat
>> >>>>>>> half truths until they make the voters believe them.
>> >>>>>>>
>> >>>>>>> John (011)
>> >>> --
>> >>> Patrick Durusau
>> >>> patrick@xxxxxxxxxxx
>> >>> Chair, V1 - US TAG to JTC 1/SC 34
>> >>> Convener, JTC 1/SC 34/WG 3 (Topic Maps)
>> >>> Editor, OpenDocument Format TC (OASIS), Project Editor ISO/IEC
>> 26300
>> >>> Co-Editor, ISO/IEC 13250-1, 13250-5 (Topic Maps)
>> >>>
>> >>> Another Word For It (blog): http://tm.durusau.net
>> >>> Homepage: http://www.durusau.net
>> >>> Twitter: patrickDurusau (012)
>> >> =============================================================
>> >> doug foxvog doug@xxxxxxxxxx http://ProgressiveAustin.org (013)
=============================================================
doug foxvog doug@xxxxxxxxxx http://ProgressiveAustin.org (014)
"I speak as an American to the leaders of my own nation. The great
initiative in this war is ours. The initiative to stop it must be ours."
- Dr. Martin Luther King Jr.
============================================================= (015)
_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J
To Post: mailto:ontolog-forum@xxxxxxxxxxxxxxxx (016)
|