Yeah, Chuck, I’m a little like you.  My wife thinks I’ve abandoned her.  But programming takes up time.  So now that the toy is written, there’s more time for the family. 

I’m convinced that if the bot doesn’t have something to say, it isn’t much better than a big knowledge base hole in which you dump facts (or opinions.)  I think this is why many chat bots fail.  Once a bot has a world in which to function, then using hints and suggestions and asking questions (which are subtle commands in disguise) gives depth to the inputs.  This is how children work.  They create their world view and then share it.  The words become attached to “real” things they believe.  Which is where I think your vocabulary experiments will lead you eventually.  You’ll find some common ground with the bot to explore.

Gary,
I agree…a bot ‘may’ not have anything to say.  I kinda like that.  This would be important when we start adding speech recognition and voice to our chat bots.  Can you imagine coming home after a long day and your bot begins yacking..blah blah blah! Maybe that’s the Loebner prize metric…telling your chat bot to ‘shut up’. =)

*Update*
I’ve been hammering out some code for my grammar parser. I hope to wrap it up so I can post it to share with others. I’m creating a ‘free form’ parser so others can customize the parsing sequence and tags used to represent the parts of speech. Additionally, it will work with ‘left to right’ languages only. It uses word order.

The parser is an Excel app so it can be used with version 2000 and up.

Regards,
Chuck

Hi,
I haven’t fallen off the face of the earth.  Took a very long weekend off with my wife in the Appalachians.  From our conversations, I’ve agreed to build here a new data model for her business.  I need to learn to use SSL, Asp.net, security, MS SQL, and other minutia related to web site development.

The downside is that this will take away my time to complete this Excel grammar parser. The upside is this new found knowledge that will allow me to move Walter and my chat bot experiments to the web.  I needed to learn this eventually. Might as well kill two birds with one stone…as they say.

Another bright spot is that my department at work needs a “knowledge warehouse”.  I’ll be supporting the design team with coding. The interesting aspect is the code will need to ‘mine data’, determine context…and in addition the idea of a program that ‘chats’ with a user to help them locate information is of significant interest.

Sure wish I could work full time on the bot experiments. 

A competing distraction is another game programming competition. It just started. If I’m sensible I’ll stay away. 

Regards,
Chuck

Ok,

I apologize for trying to sway the direction of your project.  A couple of things come to mind.  One approach that I’ve read about is to find the concept or objects of an input first, to set the context.  Then using these, fill in the specifics of the input, but doing that in combination to what the bot already knows about the objects.  That way inputs don’t have to be complete, they are more or less hints to what the bot is already doing and is aware of.  This only applies grammar in a limited sense.

What many chat bot authors try first to do is create a translator.  One that is able to represent almost anything in an internal language.  They are good for reversing the process to spit back out, maybe in another language, what was originally entered.  For some strange reason, programmers feel this is the way to make a machine do what people do.  Only that is probably not how it really works.  So heading down this path, as history has shown, is a hard row to hoe.

The results you might achieve include being able to tell your bot what is going on in its world.  Having a world defined as data is something like having an encyclopedia on a particular country.  You’ll be able to put facts into that encyclopedia and retrieve them too.  For a small part, the world database will be able to transform on its own, like the sun rises and sun sets.  Academically it might be interesting, for awhile, to investigate the virtual world.

Soon it will not be that exciting to say “the elephant wears pajamas” and then ask “who wears pajamas?”  I mean you can substitute the things in your little chat bot’s environment, but the outcome probably will be similar.  It may reach the level of being like the SIMs. 

On the other hand you can start with building a character with a personality that you find compelling. A bot having such a dimension does not need to have inputs to respond to.  It already has the hopes and desires, the situations brought about by dynamic plot twists and turns, the deep concerns and beliefs that inspire good discourse.  Then your inputs with the bot can seek common ground - maybe the challenges that are paralleled in both your world and the bot’s.

The point is, it is not the details of the bot’s world, it is the way the bot deals with them (with your help) that makes a better chat, in my opinion.  Your bot only needs to understand in terms of its own goals and intentions.  It doesn’t need to understand the universal standard for all communication (which it still then needs to integrate with its personality.)

Enough of my soap box.  There is more than one author attempting to write a bot by building a translator or search engine.  It is a popular notion. I just prefer to deal with the biggest bang for the buck first.  A bot needs a personality.

@Gary

I agree with you that a chatbot should exhibit a personality, and that’s a primary goal for me, with Morti. But a chatbot also needs to have at least a fundamental means of understanding what is being “said” to it, first. And this is what (in my opinion, at least) Chuck and Victor are trying to accomplish at this stage of their respective projects.

@Chuck

I think it’s safe to say that we’ve ALL had the “pleasure” of dealing with these types of delays, and it sounds like what you’re doing for work will net you some real benefits for Walter. Good stuff, there!

By the way, I’m currently on vacation in Tullahoma, Tennessee, and will be here for about a week. I know I’m not all that far from you, Chuck, and it’s a shame that I don’t have the means to take a side-trip over to your neck of the woods. I’ll make an effort to do so on my next trip out here.

Unfortunately, this vacation also means that Morti has also been left to his own devices, and he’s not handling it well. For some unknown reason, it seems that he’s slipped into a catatonic state that I can neither correct, nor even investigate from here; so I can see that I’ll have my work cut out for me when I get back home. I swear, you can’t leave that little rascal alone for a second!

While on vacation, I won’t be able to stop in as often as I would like to, but I’ll try to stop in when I can. I hope everyone is doing well.

Sorry to have interferred.  I saw hope in Chuck’s approach of his bot having a life of its own in a world made just for the bot.  I see now that he’s changed to the point of view where the bot is in Chuck’s world, obeying Chuck’s concepts and language. Chuck wants to talk to his bot first, before it can understand what he is saying. That way Chuck can teach his bot, somehow.  I suppose that if the things Chuck wants to talk about are not part of the bot’s world, he can just add them.

Anyway, I feel guilty for hijacking his thread.  So I’ll quit the lecturing. Is there ethics in giving a chat bot rights to live its own life as it wishes or are all these imaginary creatures ours to do with as we please?

Update
=====

Hi Everyone,
I want to apologize for my absence over the past 6 weeks. I really enjoy conversing with like-minded people who are interested in chat bots.  I really miss being here.  I must admit I have done no work on my bot. I’ve been distracted by three significant things. 

I had to take measures to get my health in focus: physical, testing, adjusting meds, exercise (yuck), dieting (super yuck), etc.  Anyhow, I’m feeling much, much better now…I simply need to continue this new regimen for the rest of my life.  =)

I had to build a workable greenhouse and a pergola for my wife (overdue promise) for our garden. Both are operational and my wife is very, very happy.

I’ve been focusing on a new job opportunity…that would move me from programmer to a training manager.  I’ll know soon if my preparation over the past several weeks and the interview yesterday paid off.  I would like to get away from professional programming. It zaps me of all my programming energy at night and on the weekends.  This new opportunity would allow me to return to my favorite hobby of programming.

Gary,

I see now that he’s changed to the point of view where the bot is in Chuck’s world, obeying Chuck’s concepts and language.

That’s an interesting way of putting it.

XBot’s (Walter) world is virtual.  It will be a dynamic environment. The reason for synchronizing XBot’s world with our world is to give XBot and the human chatting a common basis for conversation (e.g. let’s talk weather).  A judge could easily determine that he or she is speaking with a bot simply by asking a few questions about recent and local weather. If there was some other criteria for passing the Turing test then I would be inclined to have XBot’s world ‘define’ itself…allowing natural (virtual of course) variations, entropy, and other effects to run their course.

The interesting bit Gary is I did consider introducing things into the world that made absolutely no sense to people…fantasy type stuff. I thought about giving it a basic ability to ‘invent’ language. The burden would then have been placed upon me to learn it’s language. However, it didn’t take long to realize that I’m just not that smart. =)

Here’s what I’ve been doing in my spare time that could help with XBot.
* Studying very, very basic Mandarin. My purpose is to consider how a ‘universal’ grammar parser might be designed.
* Relearning my German…again to support the grammar parser.

More later.

Regards,
Chuck

Welcome back Chuck! To be fair, yes I’ve missed you!!!

Personally, I’ve had a lot of focus on my health as well. I’ve went through a detox week in Thailand which was not only very good for my health, but also very instructive. Now, back home, all my days start with this one hour ritual:
-drinking appelcider-honey tea. Very good to kick digestion, a good start of the day
-making fruit punch with a special machine. I use combinations of Fresh apples, oranges, pears, kiwi, ginger. Tastes VERY well and is extremely healthy
-15 min yoga exercise (‘sunrise greeting’)
-15 min tai chi exercise
-15 minutes meditation

Furthermore, I’ve started to swim twice a week after my wrist operation in April. Now, I’m experience the benefits of excercise my whole body (I did a lot of fitness before).

At last, I’ve reconsidered my working space to optimize my posture behind my desk.

Anyway, welcome back!

About Mandarin: : if you’d like to be connected to my chatbot contact in China, please let me know. I’ll be happy to help.

Hi Chuck, welcome back! It sounds like you’ve had a lot on your plate. I know the feeling, I’ve been getting caught up in other (work related) projects myself. Do keep us updated as you get the time to work on Walter.

I think a “universal grammar parser” is a big bite to chew. But I admire your eagerness for language learning at any rate. I need to work on my german myself—I moved to Germany this month! Any langauge-learning words of wisdom greatly appreciated. (Anyone know of bots out there designed to teach German?)

Hi Chuck. I hope that required dieting allows for the occasional sinning, otherwise, where’s the fun, he. On the bright side, you’ll probably be looking lean and mean in no time if you combine it with those exercises.  Best of luck with that new job opportunity, by the way.

@Erwin: Sounds like you had some sort of RSI problems on your wrist? I’m not certain if you are aware of this, but I can strongly suggest a kinesis keyboard. About 10 years ago, I was about to give up on using computers all together. This keyboard saved my from that faith. Even more, On a good day, I’m be able to type a lot faster with it then on a regular keyboard. 1 word of caution though: people will start to look strange at you once you start to carry the thing around all over the place

@Hunt: Is it snowing like crazy over there in Germany as well?

@Jan: did the RSI thing as well, 10 years ago. This was somehting different. But thanks for the suggestion! It keeps on requiring attention.

Hello Chuck

I’m just happy to hear such fine ideas as your last post. I recognize and appreciate exactly all that hard work you have made.

Raymond

Yup, it sure is Jan. Not so bad where I live, but further south, where I’m headed this weekend, apparently got hit hard. Speaking of getting into shape, biking to work has become something of a cross-country skiing event!

Chuck Bolin - Oct 13, 2010:

Maybe that’s the Loebner prize metric…telling your chat bot to ‘shut up’. =)

This is one of the first commands I made my bots respond to It’s essential for feedback, very useful in fine-tuning which agents respond to input ...

Chuck Bolin - Dec 2, 2010:

The reason for synchronizing XBot’s world with our world is to give XBot and the human chatting a common basis for conversation (e.g. let’s talk weather).  A judge could easily determine that he or she is speaking with a bot simply by asking a few questions about recent and local weather.

Another way is to use an agent that looks up weather on the internet. The irc bot phenny (inamidst.com/phenny) has a “.weather” command, for example:

> .weather New York
Clear ☼, 32.0℉ (0℃), 30.04in (1014mb), Light breeze 4kt (↑) - KTEB 5:51, 1051Z

So I could call that agent from my bot, then parse the results and have the bot devise some sort of relevant comment about the weather.

However, internet connections aren’t allowed in the Loebner Prize, so I would have to have the bot look up the information before the contest and store it, or something

@C R Hunt:  My bot uses Google to speak German http://www.droidAI.com

@Chuck: I think that dieting and excercising gets easier as you get stronger.
As a quick and dirty casual discussion and opinion, SSL, and security may be learned as a separate topic, like an add-on.  ASP.net is fairly easy, but I think the certification route tends to hide a reverse learning curve, perhaps to make money. 

I personally see no need to completely learn VB.NET and C# first before ASP.NET in my opinion, just to get started with basic ASP.net webpages that contain so little .NET in the code behind, especially since C# looks a lot like C++ anyway. 

MSSQL is not bad if you know SQL already, but FYI: ASP.NET works good with MySQL too.  And ASP.NET 2.0 runs C# on Linux call Mono.  There are fun alternatives to get started learning .NET web development, besides the certification route.

@Everyone: Here is my latest A.I. experiment: http://www.telecastbot.com
  It is an active experiment, so if not fully functional, consider that normal.