Going through the logs and the synthesized responses today, “teaching” corrected responses and adding new “seeds”. The slightly “gruff” “curmudgeon’ly” personality which was the goal is actually taking shape in some cases. The learning algorithms have produced some interesting successes, and some epic fails. RICH now has the references to Mistuku and Morti. (I finally gave up trying to put the link in the XML and get it past everything thats supposed to prevent that from happening LOL)  Thanks again to Steve and Dave for the referrals. Seeing the references to other AI(s) in the logs, and having just completed the Interface to the Personality Forge API, sparked a thought for an interesting experiment. A new class of interogs for AI socialization by linking through API’s.

[this is a pseudo log]

User: Have you spoken with Mitsuku recently?
RICH: Yes, just this morning.
User: What did she have to say?
RICH: [recites interaction]

or real time

User: Ask Morti what he thinks about [subject]
RICH: He said [recites response]

We would have to discuss and agree upon an API format , maybe use Benji(s) [Personality Forge] API as a basis, it seems to be solid. In addition to the sommunication standard, maybe a class of interrogs that each instance [must][should] be able to respond to. I seem to remember seeing some reference to something like this, any reference material online, anyone already working on it? Any RFC’s?

Thoughts???

Vince

I have no problem with the idea at all, Vince. If you want to shoot me an email, I can give you connection details and expected result format for Morti. I have to warn you though, that Morti’s “API” (if you want to call it that) isn’t what you would call “standardized” at all.

And not to sound ignorant, but… :RFC”???

Cool Dave, Ill shoot you an email.
RFC= Request For Comment
Basically anythig that will function over the Internet that will or should have a common format goes through this process: http://www.ietf.org/rfc.html Lets say you and I come up with a [whatever] and want to see it implemented. We write an initial paper using a particular format,  http://tools.ietf.org/tools/idnits/ and its published. The Internet and related sommmunities are then invited to comment on our proposal. The idea is then adopted, rejected, or otherwise folded spindled and mutiliated. If its adopted it may supercede prior versions of the same standard, and the whole thing is maintained by the IETF http://www.ietf.org/rfc/rfc3935.txt If you ever want to know the nuts and bolts of how anything is supposed to work in order to insure that your [doohickie] will peacefully co-exist with someone elses [doohickie], check the RFC(s)

The bot API is something that (eventually) might call for an RFC

Vince

Great! thanks for the clarification. Always good to learn new things.

> http://www.meta-guide.com/home/open-chatbot-standards

Dave has asked me to weigh-in on the standards issue.  I’m really not attached to either what the standards are, or how they are arrived at.  The important thing for me is that they work, and that people actually use them. 

Obviously the easier something is then the more people will use them.  The question is, how do chatbots talk to each other?  In the past XMPP (instant messaging, formerly called Jabber) was the main way for chatbots to talk to each other.  This may change in the new era of the “programmable web”.  It does appear that APIs are on the ascendancy, and there have been indications of a decline in the use of XMPP.

So far, there has been a marked lack of API integration platforms.  Recently, I’ve come across “Webshell - The API of APIs”, a “scripting interface for fast APIs implementation and mashup development, on a cloud platform with tools and IDE”.

> http://www.webshell.io

Also recently, Richard Wallace mentioned to me that Karsten Fluegge’s (unfortunately named) “Voice Actions” API was another way of integrating a variety of major APIs into one’s chatbot.

> http://www.mashape.com/pannous