Iyouboushi's Home

Version 1.5.4 is now available for download!

For some reason I forgot to release version 1.5.3 so this post will serve as a way to talk about what I did in that version as well.

For version 1.5.3 I did the following:

• Added the USER LEVEL command: !haiku This command will cause the bot to attempt to create a haiku using the 5/7/5 syllable structure. It’s not perfect and the counts will often be a little off but it’s a start. Needs the LB Brain to be on in order to work properly.
• Changed the “Answer Search” code slightly so that it tries to find more than one answer to the same line. This may or may not help/improve anything.
• Changed the LB brain so that it stores the starting lines in memory instead of loading them every single time something is added to the brain file.
• Changed the way the filter works slightly so that when it adds things to the LB and basic brains it can run one of the same functions. This reduces code a little and makes it cleaner.
• Fixed the Twitter aspect so that it works with the 1.1 OATH system.

It wasn’t a lot, but it was just minor improvements to try to improve the way the bot works (especially loading all the starting lines, as a huge brain file makes the bot pretty slow).

Version 1.5.4 was simply another fix to Twitter, as I was having some issues with it working reliably.  It now uses the CoreTweets.dll which is soooooo much nicer.

As always, the source code is a horrible mess.  Kids, don’t use this project as a way to actually learn how to program or hold it up to some high standard.  I’m honestly surprised this bucket of bolts even works with the way I’ve jerry-rigged it all together.  Having said that, it does work so feel free to download and enjoy.

DOWNLOAD
Binaries
Source

The awfulness continues!

I had Kon the C# Chatterbot write some more lyrics and put it with some more awful music. This one is a 58 second long metal song entitled Jolly Roger Military.

Good luck understanding SHOUTY (the name of the voice) in this one.

What happens when a C# Chatterbot writes a song? Well, I’ll tell you this: it isn’t a masterpiece!

Using my C# Chatterbot, Kon, to write all the lyrics and the tone of the song, I have created an awful 52 second-long “rap” entitled: Yo Dawg! Click here to download it

I would say “enjoy!” but I’m not sure that’s possible. XD

Version 1.5.2 is now available for download!

As mentioned, I’ve added the auto perform stuff to this version. Since making that preview post, I’ve actually added a little more to it. The commands recognized are: JOIN, MSG, SAY, RAW and WAIT. Also, if you use MYBOTNAME (in all caps like that) it will replace it with the bot’s current nick on the server.

Other things added/updated/fixed in this version: The bot can now recognize multiple admins in the config file (just add more <admin> than one like you do with <user>), two new admin level commands (!msg and !raw), the filter was updated a little, and I fixed two bugs relating to RandomTalk.

Overall I’m pretty proud of this version. Hopefully Kon will continue to improve.

DOWNLOAD
Binaries
Source

Hey everyone!

Kon version 1.5.1 is now up and ready for anyone who wants to download it. This version doesn’t add a whole lot but it fixes an issue with the ident and Real Name of the bot. The only other thing it does is add a /MSG command that allows the bot owner to message a non-channel (i.e. Chanserv, Nickserv, other users, stuff like that).

Kon seems to be attracting a lot more attention lately. I first just want to say thanks to everyone who has downloaded and tried it out. Thanks to those who sent me email with suggestions on what to do with the bot. I do appreciate it. With that said, I just wanted to say real quick that the IRC side of Kon is pretty lacking. When I first wrote the bot I never imagined it to be on par with something like mIRC or any other full IRC client. It was really just programmed to be a chatterbot. So please keep that in mind when you find that the bot is lacking something on the IRC side (like being able to auto-perform commands). I’ll add what I can when I have the time.

In the next version I’d like to add some kind of perform that will let the bot automatically join a bunch of channels and perform commands (like identifying to chanserv and nickserv). I don’t know when the next version will come though, so please be patient. :)

With all that said, here’s the links:

DOWNLOAD
Binaries
Source

I have decided to go ahead and release what I’ve been working on with Kon. This version is similar to the last version of Kon Offline that I released in that it combines the two lb brain files back into one file and utilizes kon.brain for its “Answer Search”.  For the change log for this version and the download links, continue after the cut.

Read the rest of this entry »

I’m currently working on Kon version 1.5.0 and Kon Offline version 1.2.0.

I’ve redone the way the LB Brain stores things drastically. Before it was split–questions in one file, everything else in another. That’s no longer the case. It’s all back down to one lb.brain file. Also, I’m working on something I call “Answer Search.” This uses the kon.brain file. Basically both files are now required to make Kon work (if you use that new function).

Take this example conversation we had with Kon last night:

[00:34] Kon: My skills with a sword are highly venerated.
[00:34] Iyouboushi: lol
[00:34] Smz: kon: Too bad they’re all fabricated
[00:34] Smz: XD
[00:34] Kon: I can’t rest ’til you’ve been exterminated.
[00:34] Smz: kon: Then perhaps you should switch to decaffeinated
[00:34] Kon: I’ll leave you devastated, mutilated, and perforated!
[00:34] Smz: LOL
[00:35] Smz: kon: Your odor alone makes me aggravated, agitated, and infuriated
[00:35] Iyouboushi: XD
[00:35] Kon: Your odour alone makes me aggravated, agitated, and infuriated.
[00:35] Smz: XD
[00:35] Smz: kon: You lose!
[00:35] Kon: That was too close You were almost a Jill sandwich.

As you can see from the quote in the above post, it sort of works. Kon was able to continue insult sword fighting for a few lines before it started to parrot.

Unfortunately I foresee two problems with this.

#1. It won’t be quite as random. In order to make the bot more coherent and appear smarter, the randomness has to be toned down. It’s still more random than using AIML though.

#2. As the brain grows, the “answers” it’ll find and give back to the user are probably going to be a jumbled mess. But time will tell. Right now I’m testing on a brain that contains 3000 lines. That’s pretty small considering the last brian, before I wiped it, contained 300,000 lines or so.

More testing is needed and I’m sure I’ll have to wipe it out and try a few more things before it’s all said and done.

The release date on both of these are still up in the air. I need to test more and fine-tune before I decide to let it loose upon the public. But if you can’t wait to try it out, I’ll keep Kon in #Kyoto on irc.esper.net. Feel free to come chat with him there.