Beefy Boxes and Bandwidth Generously Provided by pair Networks
more useful options
 
PerlMonks  

Re: Challenge: Designing A Computer Opponent

by Hercynium (Hermit)
on Dec 13, 2007 at 03:20 UTC ( #656742=note: print w/ replies, xml ) Need Help??


in reply to Challenge: Designing A Computer Opponent

Discussion on the CB led me to some ideas for you...

Do you use a dictionary of words that the computer opponent uses to find words to guess? If so, you could simply generate different dictionaries for different difficulty levels. There may even already be pre-computed dictionaries available online with words rated by how common/simple they are, and thus provide a convenient basis for categorization!

Also, the type of analysis required to solve the problem lends itself well to recursive solutions... perhaps build-in a (configurable) limit to how deep the analysis goes?

If all else fails, or isn't enough, I am partial to my earlier suggestion - make the computer make mistakes! Randomly inject bad data into the algorithm in the form of bad or mis-spelled dictionary entries, randomly delete previously-guessed entries, etc...

Update: While I don't have the time right now to come up with code, it looks like what's needed is some degree of intelligence for how to hamper a solving algorithm in a somewhat-consistient human-like fashion. My experience with programming for work points me towards a more straight-forward modeling approach with a pile of configurable-options that are tweaked until it 'feels' right, but my inner nerd wants something more like AI. I'll keep thinking about it, but I'd bet soup-to-nuts there's about 1000 monks here who could do this 10x faster and better than I :)


Comment on Re: Challenge: Designing A Computer Opponent
Re^2: Challenge: Designing A Computer Opponent
by Limbic~Region (Chancellor) on Dec 13, 2007 at 03:29 UTC
    Hercynium,
    Do you use a dictionary of words that the computer opponent uses to find words to guess? If so, you could simply generate different dictionaries for different difficulty levels.

    Yes. As I indicated in my original post, known vocabulary is one of the factors that can be used for a configurable difficulty level. The problem is finding the right balance.

    Also, the type of analysis required to solve the problem lends itself well to recursive solutions... perhaps build-in a (configurable) limit to how deep the analysis goes?

    I am not sure how to apply this. My current expert solution is not recursive and I am not sure how I could stop it midstream. Did you have an implementation in mind?

    If all else fails, or isn't enough, I am partial to my earlier suggestion - make the computer make mistakes! Randomly inject bad data into the algorithm in the form of bad or mis-spelled dictionary entries, randomly delete previously-guessed entries, etc...

    As I indicated in my post, mistakes are factors that can be used in the difficulty factor. The problem is an actual implementation that results in a realistic player and not something too easy or too hard.

    Cheers - L~R

Log In?
Username:
Password:

What's my password?
Create A New User
Node Status?
node history
Node Type: note [id://656742]
help
Chatterbox?
and the web crawler heard nothing...

How do I use this? | Other CB clients
Other Users?
Others studying the Monastery: (16)
As of 2014-07-22 15:33 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    My favorite superfluous repetitious redundant duplicative phrase is:









    Results (117 votes), past polls