AIML 2.0 can use sets and maps to cover those cases.

AIML 1 relies on srai to group words with the same meaning. So “wouldn’t”, “couldn’t” and “didn’t” could all match the same pattern. However, it was just a throwaway piece of code and I didn’t spend too much time on it, as I looked at the webpage with 100s of examples, which I think you posted and postponed trying to cover them all. It was just too time consuming.

Yes, I coded it to know that X was big and Y was small. It doesn’t really need the “because it was Z”, as the information is in the first part of the sentence. If something doesn’t fit into something else, it is because it is too big. The extra information is superfluous.

AIML 2.0 can, Don, but it’s not yet fully implemented, except for Program AB (and possibly the higher tiered paid services of Pandorabots).

[edit]
Looks as if I missed your answer, Steve. That’ll teach me to ignore “next page” posts.

I wouldn’t code much pattern matching to cover Winograd Schemas either, as there is a limitless variety of relationships imaginable between objects (Although that has yet to be proven in practice). Because they are susceptible to guesswork and human confusion, Winograd Schemas are not effective as Turing Test questions. They are hard to compose and only useful in large numbers.

Are the transcripts available somewhere or are they going to be?

They will be, probably next week at http://www.aisb.org.uk/events/loebner-prize . The entrants have first been asked if they have objections to their publishing. Not necessary but quite considerate.
I see Chip Vivant also entered this year. I’ll be reading.

Nicole’s results were disappointing to say the least.  Congratulations to Rose, Izar, Mitsuku, Uberbot.

I want to thank everyone who showed the patience of Job in answering my questions regarding the Loebner protocol, once I figured it out it was very simple, but I certainly had a difficult time understanding the simple.

However, I am looking forward to next year and hopefully with a better survival rate than this year.

OliverL - Sep 26, 2014:

Are the transcripts available somewhere or are they going to be?

Transcripts are available on this page (“Transcripts of each conversation can be accessed by clicking on the name of each entry.”)

I don’t know who made Trollbot but the transcript for it made me laugh.

I just read the transcript from Trollbot and I am laughing as well.  After reading my own transcript and being disappointed in the results, I found Trollbot to be refreshing and funny.  At the same time I would never have the courage to present such an entry, so to whoever thank you for the comic relief!

I bet nobody expected Trollbot!

TrollBot is a very very good imitation of a troll. In fact, I think that nobody can make the difference between TrollBot and the reals humans trolls that we can see in some forums.

Therefore TrollBot has passed the Turing test with success.

3 Cheers to Trollbot! I think it is the equivalent of a standup comedy routine, or that dog puppet that insults people, or many comedians like Don Rickles. (sp?)

And here I was about to apologise for Arckon’s manners at question 4. Well, never mind that then

For those interested to know how a logic-based A.I. (tries to) tackle these questions, I reveal my methods in the blog I recently started.

Steve - I just noticed our bots both answered the “how many letters..” question incorrectly.  If I remember correctly, we shared a bit code on that one.  Whoops! We both are total failures when it comes to testing   I think a rewrite on the <explode> syntax is in order!  Does the compile for the Loebner fully support AIML 2.0?

I found the source of the spelling count error and the explode tag is working perfectly. In the normal.txt file L A is substituted for LA (presumably due to Los Angeles?) so the code was looking for how many letters in P E R A M B U LA T E which is made up of 10 units instead of 11…

Just bad luck on the choice of word the judges chose.

Yes the compile fully accepts all of AIML 2.0 features. Dr Wallace entered ALICE 2.0 using it which included sets, maps the new wildcards and so on.