The first list is called "Mouse" - the list was achieved by finding
non-obscure [ie concepts I was able to recognise without resorting to
the definition] concepts at a 'depth' of 1 (whole link) from Mouse. I
then added some mouse-related nodes, and a few 'curious' ones [cf
'hamster'[], 'reference'[].

The list is (30 Concepts):

Graphical User Interface
GUI
icon
Human Interface Device
HID
Douglas Engelbart
input device
pointing device
mouse
keyboard
point
click
drag
double-click
right-click
menu
pointer
context-sensitive menu
touchpad
trackpad
Trackpoint
trackball
hamster
tailless mouse
laptop
notebook
mouse pad
mouse mat
optical mouse
reference

The generated graphs are:

"Mouse" FOLDOC Concept Map (Matched Ontology)
"Mouse" FOLDOC Concept Map (Detected [level 4] Ontology)


The next list is "Address" - generated by choosing non-obscure nodes
from (only) the set of nodes found at a 'distance' from Address such
that there would be 50 nodes output.

The list is (24 Concepts)

address
CPU
MMU
Memory Management Unit
register
memory location
main memory
word
address bus
program counter
pointer
type
object identifier
cache
e-mail address
ISP
Internet address
IP address
address resolution
dot notation
dot address
ARP
Address Resolution Protocol
OID

"Address" FOLDOC Concept Map (Matched Ontology)
"Address" FOLDOC Concept Map (Detected [level 4] Ontology)

The next is "object oriented programming" - generated in the same way as for
"Address". Except, "dynamic programming" is thrown in as it has nothing
directly connected to OOP.

The list is (18 Concepts):

Java
dynamic programming
Smalltalk
type
class
instantiation
polymorphism
object-oriented programming
reuse
abstract class
inheritance
dynamic binding
programming fluid
instance variable
object
multiple inheritance
concrete class
subtype

"OOP" FOLDOC Concept Map (Matched Ontology)
"OOP" FOLDOC Concept Map (Detected [level 4] Ontology)

I though a lot of these could be construed as biased, so I worked
forwards for "computer" - hand-drawing an heirarchy of computer
components and trying it out. There was an issue: often in our
ontologies nodes are treated [by the generation algorithms] as a
single concept [eg acronyms and their expansions] but the 'strict
concept mapping' methodology [ie nodes restricted to a set] may
leave gaps if only one of the concepts [eg 'ADT' but not 'Abstract
Data Type'] is allowed. So I generated a bunch of graphs, by
trying out the 'model expansion' algorithms.. then there was
another problem: the 'computer' heirarchy covers an awful lot of
the ontology [as you'd expect], such that any significant distance
will generate a very large model. So small distances and a minimum
peerage of 2 [ie no solitary concepts] were enforced.

The original list [hand drawn by myself (I'll scan later)] was:

computer
storage
disc
tape
"hard drive"
CD
"Compact Disc"
"floppy disk"
input
processing
output
HID
Human Interface Device
joystick
keyboard
pointing device
mouse
stylus
trackball
trackpad
hamster
motherboard
CPU
cache
register
instruction register
return address
jump
RAM
ROM
bus
PCI
ISA
AGP
sound card
speaker
monitor
LCD
CRT
plasma
shadow mask


In the generated graphs, there was a big difference between models
generated from the matched and the detected ontologies.

For example:

"computer" Culled FOLDOC Concept Map (Matched Ontology) Distance=0.07, Min Peerage=2
"computer" Culled FOLDOC Concept Map (Detected [level 4] Ontology) Distance=0.07, Min Peerage=2

Other 'interesting' models were:

"computer" Culled FOLDOC Concept Map (Matched Ontology) Distance=0.12, Min Peerage=2
"computer" Culled FOLDOC Concept Map (Detected [level 4] Ontology) Distance=0.08, Min Peerage=3

Note that each of these generated models has a different concept
list due to distance/culling [all the others are depth=0, Min Peerage=0]



[] hamster = a foot-controlled mouse
[] mouse -> point -> pointer -> reference
        ^\___________/^