SRL Archive - Senior News and General

Well, Seeing as cheating has turned into skilling on a single player (and caring about the character) instead of using armies (that are cared little about), I propose we create a huge auto response database, that scripts can request responses from.

Say for example, somebody asks how old a player is whilst macroing, our system would pickup the '?' or a key phrase, and send it to our database. The database then checks for possible responses and sends back a response dynamically.

We'd call it in the findfastrandoms procedure or something?

It (I believe) has been suggested before, and at the moment, I'm very up for making it. I would be willing to stay up for hours inputting responses to pre-set questions.

What does everyone else think?

The point has been made that rs will be able to have access to our srl library. Since there is no randomness in the phrases and there will only be, oh 13 and thats pushing it, different responces, they can simply put those exact phrases in as "hotkeys" and whenever they have a chat log in with that phrase, they wil monitor that account easier. It will have to be very general things or some sort of dynamic responder.

From: pwnaz0r
The point has been made that rs will be able to have access to our srl library. Since there is no randomness in the phrases and there will only be, oh 13 and thats pushing it, different responces, they can simply put those exact phrases in as "hotkeys" and whenever they have a chat log in with that phrase, they wil monitor that account easier. It will have to be very general things or some sort of dynamic responder.

Jagex won't have access to our online database of phrases…

That's what I was meaning. We get the phrases dynamically from the database and they're updated weekly. Something like that.

If you want to do the updating, it's fine with me….

From: Sp0rky
Jagex won't have access to our online database of phrases…

That's what I was meaning. We get the phrases dynamically from the database and they're updated weekly. Something like that.

Yes but unless you have it closed source, it will. jagex is here and we know it. So this is like an SSRL only thing?

I've considered this already..
And I am 100% sure Its possible to generate random sentences..
We can put a HUGE auto-respond library in a plugin, so It's closed source..

I don't consider online database as an option.. Is to slow, connecting to the database for the response..

I know that @ villavu there is a big database of conversations.. We can use those to get average phrases, and so make smart keywords to get sentences out of it.. We might want to use Regex..

I'm willing to make a good response system..


The problem isn't the responding part, but its the recognizing part.

From: mastaraymond

The problem isn't the responding part, but its the recognizing part.

Is that because of the lag when checking the text? Because I already thought of that :o (by getting the string every 5 seconds or so, then comparing it).

I don't think an online database would be too slow. How slow is sending an srl randoms report? xD

I guess it could be done in a plugin, and I'd be willing to help you code it if you needed any help. I'm also not sure about ssrl only or not, just posted it here because I thought we might actually get something done from it.

Edit: Realised we're facing some majour lag issues if it's all done clientside in a plugin, aswell as individual members not being able to keep it updated/add new phrases to the library. Defeats the point.

The current auto responder system I have is fast enough for what I'm doing. It completes the text detecting in about 50 ms. I'm not sure how fast it would be searching through possibly hundreds of phrases though. How fast is Pos?

Maybe we could start by extending the dev responder?

If we want an auto-responder that is very rare to answer wrong things, we must use GetTextAtEx and have a lot of procedures/functions to make it trigger perfectly and with as little setup as possible. We should also consider saving every triggering sentence into an array along with the one who said it. That way, we could avoid answering twice to same sentence, and make it more human, by allowing it to trigger from a message almost like the other one, but still not the same.

-Knives

Take a look at my responder (also posted an updated version in members, but I found this quicker), it does exactly what you said.

We should also consider saving every triggering sentence into an array along with the one who said it. That way, we could avoid answering twice to same sentence, and make it more human, by allowing it to trigger from a message almost like the other one, but still not the same.

From: King of Knives
If we want an auto-responder that is very rare to answer wrong things, we must use GetTextAtEx and have a lot of procedures/functions to make it trigger perfectly and with as little setup as possible. We should also consider saving every triggering sentence into an array along with the one who said it. That way, we could avoid answering twice to same sentence, and make it more human, by allowing it to trigger from a message almost like the other one, but still not the same.

-Knives

My old responder did that..

Yet the hardest part is finding recognizing what they are actually saying..
Of course you can respond to keywords like "Hi,Hello, noob, mining lvl, wsup, reported".. But thats still kinda "botish"..

With SMART, we could check the locations of players; that way we could ignore players who are further away, or focus extra on players near us - 'OMG STOP FOLLOWING ME NEWB!' would sure be nifty.

Questions:
1) Can you tell if someone is following you from the client?
2) Can you tell what someone else is doing through the client?

Why I'm asking:
1) Sorta obvious, just look at the phrase above.
2) If you could tell that someone is, say, mining, three tiles away from you, you could be the one to start the convo, 'Hey, mining lvl?', etc.

The simple trick I used in autorespond.scar is a two level answer. Take for instance the responce to bot/macro. The first responce is "huh…me" whilst the second responce if bot/macro is found again is "I am not autoing"

In that way you get a "kind of" dialog, but I have never been able to lift it beyond that. We tried implementing Eliza into SCAR, but noone has yet been able to make it at this level.

This always reminds of the Script I have been wanting to make: The Runescape Friendly Guide Script. A bot that actually interacts with newbies, teaching them basic routines, bringing them to certain places at request. Its Home would be Lumbridge Castle. It would require a lot of nested dialogs-branches. Perfect for understanding responces. It would log all the text, parsing, sculpting and enhancing the database to a Humanoid level. The main objective of such a script would be interaction.

"Dont forget your Guide!"

I've never really been a fan of responders because I've never really had any characters banned. However, apparently I'm not the only one that has made a chatbase, but I had a feature that alowed you to store conversations so you wouldn't say the same thing twice and carry on different conversations with different people. I just never made a database of possible conversations and responses.

To sound like the average Runescape player all you need to do is say penis, noob (different variations), and lol (also different variations) every 10 minutes.

From: WT-Fakawi
We tried implementing Eliza into SCAR, but noone has yet been able to make it at this level.

Wow, haha, Eliza seems excellent. I guess it's what all these chat bots are based on.

Surely we could implement it into SCAR? It wouldn't be that difficult to implement?

From: Sp0rky
Wow, haha, Eliza seems excellent. I guess it's what all these chat bots are based on.

Surely we could implement it into SCAR? It wouldn't be that difficult to implement?

Wouldn't be that hard.. But its like totally "off-topic". The bot is talking about psychic problems..

I believe someone (Solemn Wishes?) did, it's at mopar's.

From: mastaraymond
Wouldn't be that hard.. But its like totally "off-topic". The bot is talking about psychic problems..

We could easily get a chatlist/response list for a bot like SmarterChild.

Unlike previous and contemporary chatter bots that mimicked human conversation (see ELIZA, PARRY and A.L.I.C.E), SmarterChild uses a combination of complex pattern recognition and simple natural language recognition to take user questions and comments and translate them into database queries.

SmarterChild displays sophisticated learning, forming opinions of people or things based on the opinions of its user base. For example, if a number of people claim that they do not like Osama Bin Laden, SmarterChild will mimic that opinion.

SmarterChild also insists that its users engage in polite conversation, demanding apologies for harsh language and, if necessary, refusing to engage in conversations with a user until they change their tone.

:D

Finding the source for it may be a problem…