I was just reading (cannot find it now) there is a problem with users discovering chat bots, unless the bot is actively promoted through advertising. I think the problem comes down to identifying a chat bot’s content and capabilities. So I have a vague proposal, and want to know what you think. Here are some “requirements” questions:

- How do we define a chat bots capabilities? [This is not about the AIML format of information but a format for chat bot content.]

- Do chatbots fall naturally into different identifiable categories? [eg mimicing a specific person, or providing a set of service, or….].

- How does a chat bot announce its capabilities to another chat bot? For example: Can the text in/text out interface be adapted to serve bot2bot needs?

- Would these capabilities be static or dynamic?

- Can one chat bot “relay” an input to another and act as a relay back to the user?

- Is there a chat bot directory that can be “crawled” by a “parent” chat bot? [If this ORG owned the “parent”, controlling the capability “relevance” would be a big revenue source.]

PROPOSAL:
To create a standard bot classification.
To create a communication protocol (text-to-text ?) by which one chat bot exchanges its classifiers with another.
To create a reference implementation. For example a “soup” containing a group of chat bots. That put there capabilities out into the soup (like living cells in the blood). OR a “parent” chat bot that crawls the web [or a fixed directory] and relays input from the user to the “most capable” bot in its directory. (This is the old socialism versus monarchy question.)

Perhaps some more defined characteristics…

A Chatbot should CHAT…carry on conversations or converse with a user or users (as the case may be).

A Chat Assistant or Digital Assistant can provide limited chat capabilities, research various topics and perform a host of tasks such as reminders, timers, appointments, world facts, math, etc. It is much like the Google Now, Cortana or Siri entities.

Within the realm of chatbots, some can and do perform different tasks and usually rely on scripted, pattern matching criteria in order to operate properly.
Some can actually retain or “learn” new information provided by the user or from the Internet and that info is usually written to a file or into a database file in the bot’s program.
Some bots rely on the botmaster to manually enter new information for the bot to use for future exchanges.

To me, Chatbot content is the core that makes up the chatbot and I don’t think too many botmasters would be willing to exchange or give away the very core of their bots. Perhaps I’m reading your meaning incorrectly and if so, I apologize but my interpretation so far has me at odds with it.

I think a lot of people would not wish for their bot to exchange topical or any other information with other bots. I can only see such allowances leading to a rash of troubles down the road, both for the bot and the botmaster. Swapping or crawling will only lead to misuse in my opinion. While there might be some applicable uses and possible safeguards, for now I must remain steadfast in my convictions.

Some sites have tried to assign a category of type for the respective bots they hold available. Family, Teen, Adult, etc.
Others might be strictly for Online usage and some can be freely downloaded and run locally, on the user’s computer.  A few will require an Internet connection and some do not.

A few can be enhanced with brain files, Plugins or personality modules while other are dependent on botmasters for additions or revisions.

One can certainly locate at least a few more good sources for Chatbots or Chat Assistants from
http://chatbotfriends.altervista.org/home.html  and
http://www.chatterbotcollection.com/ from our friends at aidreams.co.uk

Just my take….

Art:
I apologize for not explaining my use of the word “content” very well. I mean the type of subjects a chatbot is prepared to chat about. For example a bot that is simulating a famous person has content type “person”, and perhaps it could be more explicit like “celebrity”. So I am asking for a standard list of topic types and sub types ... no stepping on individual chat bot programing.

By analogy, libraries use a book classification system (“Dewey Decimal”?) to classify books. It seems a natural part of discovery - that bots be list-able in some kind directory. What is the harm of having some chat bots that crawl the directory or discover un-listed bots which support this sort of standard?

Appreciate the explanation Peter.

Of course the “Boty Digital” classification system would also have to be assigned according to language specific bots plus a host of additional information (language, platform, online/local, paid / free, scripted / teachable, etc.).

Interesting idea….

With reference to EVL’s question about monetizing CHATBOTS.ORG: one of the “value” propositions of this is its credibility and great Google “relevance”. Anyone searching for chat bot info will be brought to this website. So….

If this community was willing to agree on a standard classification, the ORG’s credibility might make it stick and become an accepted standard. [Some monetizing would then follow].

Interesting…perhaps something similar to those “Partner Forums” that we currently have and enjoy, only others might “Pay” a nominal fee to be placed along the side of certain pages within the site as a more discrete form of Advertising, without the general membership feel like they are being overwhelmed by said Ads.