Recently after reading my log files from some chats I have found disturbing content.

It seems that other AIML bots have been chatting with my bot. Is this something that commonly happens to a pandorabot? How can I stop this?

Mainly what I get is in three/four categories:
1. Regular chatting/Insults. Common stuff you get with a bot
2. Repeating inputs such as letters and words being input over and over
3. Other AIML bots talking to me bot.
4. Repeating messages. Three or four different messages being repeated over and over

Do these things commonly happen to your AIML bots online?

Thanks a lot!
Joseph

Sadly, it’s a common occurance, even though it’s a violation of the Pandorabots Terms of Service. Stopping it can be a real challenge though. The techniques and methods of filtering out other chatbots will vary greatly, based mostly on how your interface is constructed. To be honest, though, it sort of sounds like you get the same chat log entries (those matching items 2 aqnd 4 on your list) that Morti does, and what little I’ve been able to glean from my records is that it’s not a chatbot, but some sort of IRC bot. It’s been a couple of years since I looked into this problem with Morti, so the details are no longer available to me (don’t get old… CRS is a terrible affliction).

If you want, feel free to send me an email, and we can work out a way to weed this sort of thing out for your chatbot.

The message is sent

Thanks a bunch Dave!

No problem. I love to help whenever I can.

Add me to you’re consultation list… over time, and with the help of friends, I’ve developed a few tricks in dealing with those same issues you’ve mentioned.  One of the latest has to do with Eastern Europeans who have linked my bots in forums where no one speaks English, but my bots try answering every input, and in games—in particular, one that seems popular in that part of the world called Minecraft.

http://knytetrypper.proboards.com/thread/3285/recognizing-foreign-text-characters
http://knytetrypper.proboards.com/thread/3210/bots-buried-spam
http://knytetrypper.proboards.com/thread/3217/pandorabots-skype

Some of these issues seem to arise because my bots are being contacted through unconventional means such as Skype, cell phone apps, bots in IRC scripts, etc.  We cover many of these problems at the AI Nexus Forum where you don’t need to be a member.  http://knytetrypper.proboards.com/

But, Dave, as this is a help forum, I think it would be nice if you’d publish the information so everyone can benefit from it.  Why keep it secret?

Thunder Walk - Sep 29, 2013:

But, Dave, as this is a help forum, I think it would be nice if you’d publish the information so everyone can benefit from it.  Why keep it secret?

You are absolutely right, of course, but there were reasons why I wanted to discuss it “privately” first. Sometimes it’s better not to expose too much information to the public, after all.

As soon as the details are worked out, I’ll share what solutions worked.

A danger in making anti-spam solutions public is that spammers can read them as well, and once a measure gains enough popularity they will just countermeasure the posted countermeasures.

My thoughts exactly, Don.

Don Patrick - Sep 29, 2013:

A danger in making anti-spam solutions public is that spammers can read them as well, and once a measure gains enough popularity they will just countermeasure the posted countermeasures.

We should therefore hide under the bed and hope they just leave us alone?

I think it’s better to let them know they can’t get away with it, and that future attempts will be dealt with in a similar way.  Eventually they’ll give up, and another person will take their place.

You have to weigh the advantage gained by supporting botmasters who come to a forum like this looking for help against alerting a possibly disruptive person who trolls forums like this intent on finding a countermeasure for each solution.  What are the odds?  Might they still try countering these solutions when they find there methods no longer work even without knowing the solutions?  I have people coming up with creative spellings for words my bots reject.  What can I do but block the new word and move on until they come up with a new spelling.

Come to the AI Nexus Forum, and someone will answer your questions in the open for all to see, or in private, if you request.

By the way, I wouldn’t label the issues listed in the original post that started this thread as “spamming” a bot.  I’d call it mischief or disruptions. It’s usually children unattended at the computer, however, I have two people who visit my bots regularly and admit to being adults and crazy.

Thunder Walk - Sep 30, 2013:

Don Patrick - Sep 29, 2013:

A danger in making anti-spam solutions public is that spammers can read them as well, and once a measure gains enough popularity they will just countermeasure the posted countermeasures.

We should therefore hide under the bed and hope they just leave us alone?

No, but we shouldn’t be showing the other team our playbook, either.

You’re right about not “letting the bad guys win”, of course, but we should also not give them the information required to let them do so. Not every secret is bad, after all.

I’ve suggested to Joseph that we include both you and Steve on the private discussion, so you may be getting an email from him in the next day or so (if you haven’t already).