The AIML Chatbot kernel

The AIML Chatbot kernel wraps the pyAIML Python library to create a conversational bot within a notebook. The bot behaviour is defined by loading AIML categories. A "category" in AIML is akin to a rule, and defines a pattern-template pair

Input cells are messages for the bot, which get interpreted and the bot output is the result of the cell execution. Additionally, a few magics for both control (cells starting with apercent sign) are available; these are not passed to the bot as user messages, but executed to change the bot behaviour in some way.

First steps

On an empty bot there is nothing to do (the bot does not produce any output, since it does not have rules to work with). So when creating an (empty) AIML notebook, any input cell will just return the general help message



In [1]:

    
Hello









    




AIML Chatbot

You can start by loading a database of rules:

%learn alice | standard | <dbdirectory> | <xml-file>

For "alice" & "standard" databases, the rules will
automatically be activated. For a custom database,
you will need to launch the "load <name>" command
defined in it.

Once loaded, you can start chatting with the bot. 

New databases can be added by additional "%learn" commands.

Use "%lsmagic" to see all the available magics.

Getting help

This same help message can also be obtained at any time by using the %help magic command



In [2]:

    
%help









    




AIML Chatbot

You can start by loading a database of rules:

%learn alice | standard | <dbdirectory> | <xml-file>

For "alice" & "standard" databases, the rules will
automatically be activated. For a custom database,
you will need to launch the "load <name>" command
defined in it.

Once loaded, you can start chatting with the bot. 

New databases can be added by additional "%learn" commands.

Use "%lsmagic" to see all the available magics.

And the %lsmagics command lists all available magics with their meaning



In [3]:

    
%lsmagics









    




Available magics:
%aiml  %forget  %help  %learn  %load  %lsmagics  %save  %setp  %show bot  %show session  %show size

%aiml  : add additional AIML rules
%forget  : reset the bot
%help  : show general help
%learn alice | standard | <dbdirectory> | <xml-file> : learn an AIML db
%load <name> : load bot state from a file
%lsmagics  : list all magics
%save <name> : save bot state to a file
%setp [bot] <name> <value> : set a predicate, or a bot predicate
%show bot  : show the defined bot predicates
%show session  : show the predicates defined in the session
%show size  : show the number of categories loaded in the bot

Learning

The most important of these magics is %learn, which loads an AIML database into the bot.

There are three versions for that:

Learning a predefined database: use %learn alice or %learn standard. AIML categories defined in the DB will be automatically loaded.
Learning a custom database: use %learn <directory> where directory is a directory containing a set of AIML files to learn, plus an startup.xml file referencing them. A load <name> command is needed afterwards to load the learned rules into the bot
Learning a single AIML file: use %learn <file.xml> or %learn <file.aiml>

See 02-chatbot-alice.ipynb for an example of the first type.

An additional way of creating rules is by writing them directly in notebook cells, using the %aiml magic. Such rules are immediately added to the bot when the cell is executed. You can:

Start from scratch (inserting rules to an empty bot). See 03-chatbot-inline.ipynb for an example
Add rules to a bot already filled with categories. See 04-chatbot-alice-augment.ipynb for an example

Version: 1.0 (2016-07-19)
Author: Paulo Villegas