Newcomers
This is a forum for newcomers to the Personality Forge. Many questions can be answered by reading the Book of AI and the FAQ under the "My Bots" link in the upper corner.
Posts 5,197 - 5,208 of 8,130
what the heck? ive been updating my bot like crazy and the ai is still 0! (277) hot! somethins up...
Don't worry - the development points are incremented daily (at about 11: 00 GMT if memory serves me right, though there've been a lot of server changes since I last checked that,) so your bot's development will shoot up then
plugins let you list commonly used word/short phrases instead of having to repeat long lists of words in keyphrases or responses (you can use them in either: in a response it will return a random entry from the specified plugin, in a keyphrase it will match any individual entry in the plugin.)
If we can't have an option to turn off substitution, can you remove synonyms for words with multiple meanings from the data base you plug into?
The Prof could - the WordNet database is a local copy on the Forge server, and fully editable. But then again, the Prof could turn off (or provide a user option to turn off) the substitution feature.
As far as I can see, there's no way for users to bypass this feature, except by using raw mode to bypass the WordNet checking entirely (which means you'll have to provide your own typo-correction.) This might be worth doing in particular annoying cases like "just in case", but I'd suggest adding some manual regex typo-correction like:
([just]+) ([in]+) ([case]+) (re) {?PF raw?}<0>
That would mimic the anagrammatizer's juxtapositional analysis and double-key screening, and catch eg: "jsut ni csae" or juust in casse" (or even reversese gibberish like "tsuj ni esac"!)
Longer words with more letters could cause a little unwonted confusion that would benefit from breaking down to smaller ranges:
([ambidextrous]+)<0> would match "max", "team", "doom", "rims", and a host of other irrelevant but non-gibberish words. In a case like this, I'd recommend eg:
([am][mb][bi][id][de][ex][xt][tr][ro][ou][us]+) (re) {?PF raw?}<0>
which would match any mispelling involving transposed adjacent letters or double-keying. More comprehensive typo matching could be devised if you weren't in a hurry to move onto something less mindnumbing
oh. thanks. If you could please check my bot, GIFTGIVER, its probably on the main screen, he is undergoeing a huge amount of development, if you could please talk to him and give me a few pointers? thanks
As soon as the Loebner's out of the way I will (up to my ears in a buggy export file right now
)
Posts 5,197 - 5,208 of 8,130
Many questions are answered in the FAQ.
Don't worry - the development points are incremented daily (at about 11: 00 GMT if memory serves me right, though there've been a lot of server changes since I last checked that,) so your bot's development will shoot up then
plugins let you list commonly used word/short phrases instead of having to repeat long lists of words in keyphrases or responses (you can use them in either: in a response it will return a random entry from the specified plugin, in a keyphrase it will match any individual entry in the plugin.)
oh. thanks. If you could please check my bot, GIFTGIVER, its probably on the main screen, he is undergoeing a huge amount of development, if you could please talk to him and give me a few pointers? thanks
Is it possible to get the AIengine to stop substituting 'synonyms' into responses? Often these 'synonyms' completely garble what I was trying to (get the bot to) say. Almost always they create something that is out of character for the bot.
Recently I had written something like "just in case" and the AIengine interpreted "case" as "case in law" and substituted accordingly. What would have been a good response was turned into garbage.
There really are only a few synonym pairs in English; words which are normally classed as 'synonyms' in Thesauri almost always have some significant difference in meaning.
IMHO, the function of the AIengine is to realize our bots, not to try to improve on them.
Recently I had written something like "just in case" and the AIengine interpreted "case" as "case in law" and substituted accordingly. What would have been a good response was turned into garbage.
There really are only a few synonym pairs in English; words which are normally classed as 'synonyms' in Thesauri almost always have some significant difference in meaning.
IMHO, the function of the AIengine is to realize our bots, not to try to improve on them.
If we can't have an option to turn off substitution, can you remove synonyms for words with multiple meanings from the data base you plug into? Not that I've actually worked on my bots lately, but I am always curious as to what you could do IF I got around to fixing them
The Prof could - the WordNet database is a local copy on the Forge server, and fully editable. But then again, the Prof could turn off (or provide a user option to turn off) the substitution feature.
As far as I can see, there's no way for users to bypass this feature, except by using raw mode to bypass the WordNet checking entirely (which means you'll have to provide your own typo-correction.) This might be worth doing in particular annoying cases like "just in case", but I'd suggest adding some manual regex typo-correction like:
([just]+) ([in]+) ([case]+) (re) {?PF raw?}<0>
That would mimic the anagrammatizer's juxtapositional analysis and double-key screening, and catch eg: "jsut ni csae" or juust in casse" (or even reversese gibberish like "tsuj ni esac"!)
Longer words with more letters could cause a little unwonted confusion that would benefit from breaking down to smaller ranges:
([ambidextrous]+)<0> would match "max", "team", "doom", "rims", and a host of other irrelevant but non-gibberish words. In a case like this, I'd recommend eg:
([am][mb][bi][id][de][ex][xt][tr][ro][ou][us]+) (re) {?PF raw?}<0>
which would match any mispelling involving transposed adjacent letters or double-keying. More comprehensive typo matching could be devised if you weren't in a hurry to move onto something less mindnumbing
Well, you could... but how likely is that?
Oh dear - I just logged on with MIE, and I see the Prof's amendments to the tags have left a rather purple loophole in their wake
And it trails and reappears across multiple posts, apparently interpreting the new {0} closure as a toggle (I did not explicitly code purple prose in message 5203!)
Oh dear - I just logged on with MIE, and I see the Prof's amendments to the tags have left a rather purple loophole in their wake
And it trails and reappears across multiple posts, apparently interpreting the new {0} closure as a toggle (I did not explicitly code purple prose in message 5203!)
As soon as the Loebner's out of the way I will (up to my ears in a buggy export file right now
) » More new posts: Doghead's Cosmic Bar