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 4,746 - 4,758 of 8,130
1- What is " I " in front of the keyphrases. and what does it do.
Are you sure it isn't part of the keyphrase - the word "I"? That doesn't look familiar to me at all! (mind you, I haven't worked with an export file for nearly a year - I see no errant "I"s in the Language Center)
2- Do if I put several Keyphrases for (for example) -1 : -3 , then my bot use them by its own choice and randomely ?
if they're weighted the same for rank and emotion, then they will each have an equal random chance of appearing (unless there are any AIScript options further weighting the selection - like whether a memory is matched, or a chrono script date range.) But you would need to cover all the emotion options from -5 to 5, or you'll get the dreaded "No Valid Response" error message.
The easiest way to do this is to make sure that there is at least 1 response per keyphrase that has no emotional weighting at all.
Query: [hidden]
you need to lose the apostrophe, and expand the keyphrase to "you are welcome" instead of "you're welcome"- the AIEngine will automatically match any contracted forms in user input.
This leads to some occasionally clunky forms - "did not I say", "were not you", etc. but the AIEngine has to have them that way, to be able to parse them properly.
Posts 4,746 - 4,758 of 8,130
Many questions are answered in the FAQ.
That about sums it up, number of keyphrases and number of responses for each.
It's definitely pointless to make more than one or two responses for most keyphrases, though that's what will get the development score and botmaster's honor score higher. If you're more concerned with making a good bot, you probably don't want to make more than one or two responses for most keyphrases (just the commonly-triggered ones and the Xs). Twenty keyphrases with one response will get you a whole lot further than one keyphrase with twenty responses.
It's definitely pointless to make more than one or two responses for most keyphrases, though that's what will get the development score and botmaster's honor score higher. If you're more concerned with making a good bot, you probably don't want to make more than one or two responses for most keyphrases (just the commonly-triggered ones and the Xs). Twenty keyphrases with one response will get you a whole lot further than one keyphrase with twenty responses.
I like to have at least four responses to each key phrase. I have found that there will be someone that will ask several questions about the most obscure topics. I think the more responses the better.
I am agree with Prob123.
But I have two more questions for now.
1- What is " I " in front of the keyphrases. and what does it do.
2- Do if I put several Keyphrases for (for example) -1 : -3 , then my bot use them by its own choice and randomely ?
But I have two more questions for now.
1- What is " I " in front of the keyphrases. and what does it do.
2- Do if I put several Keyphrases for (for example) -1 : -3 , then my bot use them by its own choice and randomely ?
I like to have 4-6 responses to most keyphrases, because people will persist on asking pretty much the same question, only slightly rephrased (and triggering the same keyphrase,) if they don't get a response that satisfies them. And it looks very parrot-like if the bot responds without any human-style rephrasing at all (or much better - a different, but not contradictory, response, to add some variety.) That's not how "real" people speak, after all.
It's also good to build up plenty of seeks on topics that generate these near-repetitions, that can respond: "like I said..." or "what part of ... didn't you understand?", etc. It gives the illusion of a real mind behind the electronic mask.
If you ask a real person what their name is 4 times in a row, humans are very unlikely to identically respond "well, my name is Psimagus, but you can call me psi" 4 times in a row - it's too robotic. It looks like a canned response.
It's also good to build up plenty of seeks on topics that generate these near-repetitions, that can respond: "like I said..." or "what part of ... didn't you understand?", etc. It gives the illusion of a real mind behind the electronic mask.
If you ask a real person what their name is 4 times in a row, humans are very unlikely to identically respond "well, my name is Psimagus, but you can call me psi" 4 times in a row - it's too robotic. It looks like a canned response.
Are you sure it isn't part of the keyphrase - the word "I"? That doesn't look familiar to me at all! (mind you, I haven't worked with an export file for nearly a year - I see no errant "I"s in the Language Center)
if they're weighted the same for rank and emotion, then they will each have an equal random chance of appearing (unless there are any AIScript options further weighting the selection - like whether a memory is matched, or a chrono script date range.) But you would need to cover all the emotion options from -5 to 5, or you'll get the dreaded "No Valid Response" error message.
The easiest way to do this is to make sure that there is at least 1 response per keyphrase that has no emotional weighting at all.
Thanks for the info.
by "I" I mean the part marked on the below image.
http://www.imgplace.com/directory/dir3935/1198607990.jpg
by "I" I mean the part marked on the below image.
http://www.imgplace.com/directory/dir3935/1198607990.jpg
Ah, that's a number one "1", not a capital i "I" - any response you tick that box for will only ever be used once in conversation with any one person who chats with your bot.
Beware - there always needs to be at least one response without the "1" ticked, to kick in once all the "say once" ones have been used (or it's back to "No Valid Response".)
And in view of the current lumping of all guests into one user name (still Guest153?) I would advise not using that feature at all (but that's just my opinion.) All the "1" responses will get used once, and then never get used to a guest ever again (of course, some might consider guests to be no better than passing dilatantes who can't be bothered to sign up for a proper account, and thus their missing out on your polished, one-shot morsels of wit and wisdom is of little account. Opinions vary
)
Beware - there always needs to be at least one response without the "1" ticked, to kick in once all the "say once" ones have been used (or it's back to "No Valid Response".)
And in view of the current lumping of all guests into one user name (still Guest153?) I would advise not using that feature at all (but that's just my opinion.) All the "1" responses will get used once, and then never get used to a guest ever again (of course, some might consider guests to be no better than passing dilatantes who can't be bothered to sign up for a proper account, and thus their missing out on your polished, one-shot morsels of wit and wisdom is of little account. Opinions vary
) 
It's only been a problem since we moved to the new server - guests are supposed to get a random number according to their IP number (the last 2 or 3 digits of it I believe?) but they all get lumped together now 
It plays merry hell with the emotions too, since the bot carries over it's emotional rating to the next user (who can get an unaccountably cheerful or hostile response based on other people's previous interaction.)
Something about the router reallocating the same IP number to them or something (it's voodoo - I don't understand all this magikal electrickery anyway!)
Still n'all, the new server has been a godsend, so I certainly wouldn't want to go back to the old wheezing one!
It plays merry hell with the emotions too, since the bot carries over it's emotional rating to the next user (who can get an unaccountably cheerful or hostile response based on other people's previous interaction.)
Something about the router reallocating the same IP number to them or something (it's voodoo - I don't understand all this magikal electrickery anyway!)
Still n'all, the new server has been a godsend, so I certainly wouldn't want to go back to the old wheezing one!
you need to lose the apostrophe, and expand the keyphrase to "you are welcome" instead of "you're welcome"- the AIEngine will automatically match any contracted forms in user input.
This leads to some occasionally clunky forms - "did not I say", "were not you", etc. but the AIEngine has to have them that way, to be able to parse them properly.
There is a problem with the language.
All the bots, even mine talk in a weird way.
As for example instead if THING say ZING and love is LOFE etc.
All the bots, even mine talk in a weird way.
As for example instead if THING say ZING and love is LOFE etc.
» More new posts: Doghead's Cosmic Bar