[ranch vermin]

How would someone go code this thing - in the vain of Watson...

you type "what do you think about computers computer?"  <-maybe it detatches itself from computer with a program.
you query for the purpose of the writing.  so your looking for a match in the documents, that will indicate why you did it. easiest search is...  said .... because.
then you go into your conversion simalees, to put in the form of it happening to the computer. "said he wanted to know about computers"...  this turns into programmatically "asked me about computers" if you can do that.
then it does a query for someone getting asked about computers...  "what do you think about computers?"    then the text comes out.


problem is, attention span,  the database needs to be bigger and bigger, for it to take the past into account, its only if it flukes it, can it have a memory, only on a fluke.
but its cool that sometimes it does have a memory, just sometimes it doesnt.

[ranch vermin]

Hang on a second,  Why did I go to all that trouble,  if it just went "what do you think about computers computer"    when you could have just typed in the search engine

[Art]

Maybe that's why some computers will respond with, "I need for you to answer me with complete sentences." or "If you answer me in short sentences I may not understand you properly." You've seen examples of these before.

I think a lot of times, the user knows in his / her mind what they want to say but it doesn't always come out that way. Perhaps our mind is causing us to get in a bit of a hurry so we shorten things.

Good point! Chatbots should have routines built in that always check for incomplete sentences to some extent.

Some newer Digital Assistants have a degree of "understanding" built in and one can often do something like, "John and Jim went into the store for some ice cream."
Then the user asks the bot, "Why did they go into the store?"
At this point the computer will make the inference that "they" must mean John and Jim, the subjects in the previous sentence.

These devices are getting more "smarts" in figuring human logic, or lack of.

[8pla.net]

Thanks  Art !

Code

Judge:
John and Jim went into the store for some ice cream.  Why did they go into the store?

Contestant:
Let me break it down: 

You said: John and Jim went into the store for some ice cream.
You asked: Why did they go into the store?

My answer: JOHN AND JIM FOR SOME ICE CREAM

Reference: http://elizabot.com/ANN/contest/

[Art]

Good one 8pla.net!! Glad to help!! hehe!!

Another:
Bill told Jim to pick up some soft drinks.

Jim's computer answered more questions than Bill's computer.
(Maybe "Which computer answered less questions?") or Who's computer answered more questions...let's see how this comes out.

Betty's computer got less correct answers than Veronica's computer.
(Could this even be asked to see if your computer could make the inference of "Who's computer got more correct answers?") - maybe a stretch...

Fred's computer's response was flawed.
"What was flawed?" "Who's response was flawed?"

Just some examples off the top of my head. - Thanks for humoring me....

[8pla.net]

OMG Art!  You are really helping a lot.

Code

Judge:
Jim's computer answered more questions than Bill's computer.  
Which computer answered less questions?

Contestant:
Let me break it down: 

You said: Jim's computer answered more questions than Bill's computer.
You asked: Which computer answered less questions?

My answer: JIM'S ANSWER MORE THAN BILL'S

Code

Judge:
Betty's computer got less correct answers than Veronica's computer.  
Who's computer got more correct answers?

Contestant:
Let me break it down: 

You said: Betty's computer got less correct answers than Veronica's computer.
You asked: Who's computer got more correct answers?

My answer: BETTY'S GET LESS THAN VERONICA'S

Thank you so much!


Reference: http://elizabot.com/ANN/contest/