Reading from and Saying to the Internet

Revised 7/10/2002 by Seth Russell

[sailor agents]

The Perdicament:  

Internally a sailor agent interacts with the computer platform on which it is hosted and it's knowledge base to produce behaviro. Externally agents read, write, and respond to languistic statements from the internet and from their owners.

Language:

It takes three things to make a language:

1. A grammar
2. A vocabulary
3. A group of interacting agents using that grammar and vocabulary.

Several languages have been defined for sailor agents.  Each language has a file that defines it's grammar. These files reside in the /homePath/grammar/ directory.  They define the grammar for both reading and saying.  The vocabulary of sailor agents is primarialy learned by reading from the Internet and is independant of the grammar in which it is read or written.  In other words an agent can read in knowledge in one grammar and write it out in another grammar.  

Additionaly a running instance of a sailor agent will respond to instructions, called semenglish imperatives. Every public method of the sailor's running instance is a vocabulary word in the instruction language. The grammar of this language is given @here@.

Note that by convention all grammar names are lower case.   Here is a  list of  proposed grammars for agents with their current implementation status:

• semenglish (saying)
• rdf
  
• quads (reading, saying)
• n3 (partual reading)
• kif
• cycl
• html (saying)
• semenglishImparitives (reading)
  

Designating a grammar for reading:

Prior to instructing the agent to read, the grammar must be designated.  This is done with the command 'reading' which takes for its only argument the name of the grammar.   For example: if you are communicating with the agent from a HTML internet page that is served by a sailorServer and want to designate that the grammar of the next read instruction will be RDF, then the command would be:

{! reading(rdf) !}

Or from a query:

http://host:port/path?reading=rdf

Or from a python script:

me.reading('rdf')   Where me is the name of your sailor instance.

Reading commands can be nested.  To revert to the prior grammar for the next read instruction, issue the command:

me.endReading()

Designating a grammar for saying:

Prior to instructing the agent to write, the grammar must be designated.  This is done with the command 'saying' which takes for its only argument the name of the grammar.   saying commands can be nested.   For example: if you are communicating with the agent from a HTML internet page that is served by a sailorServer and want to designate that the grammar of the next write instruction will be Semenglish, then the command would be:

{! saying(semenglish) !}

However it is well to note that the sailorServer starts each page defaulted to saying 'html'.  

Or from a query:

http://host:port/?saying=semenglish&at=Seth&say=node

Note that the sailorServer will send the designated query back in the language specified prior to serving the designated page.  In the case above, there is no application page specified; so the node named 'Seth' will be returned in the language semenglish.    

Or from a python script:

me.saying('rdf')  Where me is the name of your sailor instance.

Note that by convention all grammar names are lower case.

Reading:

me.read('Url')

Or from a html form:

<form>
<input type="hidden" name="reading" value="rdf">
<input type="text" name="read" size="90" maxlength="255" value="">
<input type="submit" value="read">
<input type="hidden" name="endReading" >
</form>

saying:

me.say('production')