Microsoft Word OpenIssuesInOO. doc

5.1 

Modeling in prototype-based languages 

The modeling perspective of prototype-based languages is based on a theory of people 

identifying new phenomena by comparing their properties to phenomena they already know.  

The standard example [Lie86] is the elephant Clyde that represents a prototypical elephant 

in the mind of some person. If this person encounters a new situation with another elephant,  

say one named Fred,  facts about Clyde can be used for understanding properties about Fred. 

In general a new phenomenon is identified as being similar to one or more  

(prototypical) phenomena well known to the perceiver. 'Similar' means that the new 

phenomenon has the properties of the parent phenomenon except for certain specific 

properties that may differ, not exist, or be new. In general the description of a new 

phenomenon lists the properties that differ and for all other properties the description may 

be found at the parent phenomena. Technically, references to descriptions in parent objects 

are handled by delegation. 

The rationale behind this modeling perspective is that this is how children perceive the 

world and that people in general perceive new phenomena in this way. 

Consider some additional examples: Max the dog may be grasped by a child who 

notices properties of Max, like '4 legs', 'a tail', 'a mouth', 'sharp teeth', 'is soft', etc. The child 

then meets Kit (a cat).  Kit is like Max, except that it also has 'sharp nails' and can 'crawl in 

trees'. The child may then meet Boni (a pony) that is like Max except that it does not have 

sharp teethes, but it has 'hooves', 'mane', etc. Then Lassie (the dog) shows up. It is definitely 

like Max.  New animals horses, cows, and even the elephants Fred and Clyde may be 

perceived by this child. 

The understanding of real world phenomena using the prototypical view may be useful 

in early stages of analysis where new phenomena are encountered and their properties and 

relations to other phenomena are unknown. However, as time goes on and more and more 

properties become known, the knowledge must be structured and organised systematically - 

otherwise there will be chaos. 

The prototype-based view does not capture what is special about Fred and Clyde and 

what is general about them. What does 'like' mean? Is it size, form of teeth, male/female. 

During the knowledge acquisition process you will organise your phenomena into groups 

according to similarities. Lassie, Max and Kit don't eat grass. You realize that Lassie and 

Max are more alike than Kit, Fred and Clyde, etc. You form concepts/abstractions and make 

generalisations/specialisations of these concepts. I.e. you classify your phenomena and 

concepts. Eventually you form concepts, like Dog, Cat, Elephant, Horse, etc. You can now 

use the new understanding/knowledge in the form of concepts to get a better understanding 

of your domain. 

---