{} Agent

An instrumentality for accomplishing some task.

An Agent is related to the concepts, of instruction, operator, function and method but is more constrained.  In particular an Agent specifies how some Task might be achieved and differs from the related concepts by the specification of how its invocation comes about.  Agents identify their abilities by a pattern of Symbols called its Solution Pattern.  If Tasks may be thought of as parameter lists in search of methods the methods they seek are to be found in Agents.

Many different Agents may fit the Solution Pattern of a Task.   Meeting this criterion only establishes that an Agent can in principle perform such tasks--not that it can perform a particular Task in that Task's context; nor even that an Agent is fit to perform a matching Task.  Once a list of Agents that can perform a Task has been created, the fitness of each Agent to the Task must be measured.  This is accomplished by comparing the AttributeSets of the Task with that of prospective Agents.  The two sets are first tested for conformance.  If two sets conform then the (vector space) distance between them can be computed over a normalized range of 0.0 to 1.0 inclusive.  The Agent closest to the task is ranked as most fit.

(Zylaphon's Agents should not be confused with so called rational agents of certain AI disciplines.)