Re: $100.00 Programming Contest
David Cuny writes:
> 0. Will all words appear in Junko's dictionary?
Not necessarily, but you can count on the vast majority
of them being there, say 90% or more.
> 1. Does case matter? That is, the encrypted
> string "This is encrypted" and
> "THIS IS ENCRYPTED" are essentially the same thing.
You'll know the case from the case of letters in the input.
Your output should have the same case.
Maybe I don't understand what you mean.
> 2. Can anything that isn't a character (other than apostrophes)
> be treated as whitespace? For example:
>
> "This-is encrypted!\n\n"
>
>is basically:
>
> { "THIS", "IS", "ENCRYPTED" }
Junko's dictionary does contain some hyphenated words.
A hyphen could be considered part of a word,
but in many cases it could be considered a separator
between two words. Maybe I'll defer for now the decision about
whether to include hyphenated words in the input.
> This would disallow having to deal with apostrophes:
>
> can't
> they're
The above two words are in her dictionary as one word.
> and hyphens:
>
> reddish-blue
>
> would just become a pair of dictionary words.
"reddish-blue" is not in her dictionary, but "reddish" and "blue" are.
I guess if your program has trouble with hyphens, apostrophes etc.,
it could just ignore those words, at least initially.
> Of course, I'm assuming that you want the output formatted
> the same as the inputs.
Yes.
> 3. How many words are in the input sentances?
I don't think the length of a sentence matters very much,
but I was thinking of testing competing programs on a
variety of inputs, from one to 5 typical length sentences.
It depends on how good the programs are.
> I'm curious how the "goodness" of a particular solution
> will be evaluated. Once strings are parsed into tokens
> (fairly trivial), there are a number of
> different approaches you could take, all of which are legitimate.
> Assuming that a dozen people submit a "winning" program,
> what criteria do you then use? Points for clever recursion,
> less lines of code, speed, or being the closest to the solution
> that Robert's written? 
I was thinking of running the competing programs on
several input texts of various lengths, and deciding the winner
based on the number of correct words in the output.
I don't want to use any subjective measures of goodness.
By the way, I haven't solved this problem myself.
I started to work on it many years ago, and found it intriguing,
but I didn't get very far.
> One other thing - just because a sentance can be decrypted
> into legal words doesn't guarantee that it's correct.
> For example, 'j' and 'v' are fairly uncommon. So if the
> word was "jibe" and we translated it "vibe", is it still
> a good solution (assuming all the other words were
> 'translated' as well)?
I'll be judging based on what the input text actually was,
not on what it could possibly have been.
Kat writes:
> A question David left off: What if the program uses Karl's interpreter?
The donor requested that winning entries must
run on the standard RDS Euphoria 2.3 interpreter.
Regards,
Rob Craig
Rapid Deployment Software
http://www.RapidEuphoria.com
|
Not Categorized, Please Help
|
|
