View Full Version : Inheritance and Attributes

2008-03-07, 12:24 PM

I already suggested to think about the possibility to differentiate codes
as subject, attribute, action, process, using different types of icon and
why not, a little bit different properties.

Here are some thoughts about an attribute concept. I'll describe
two very different way about how to implement it.

First way: Attribute Code

a) Such a code would be only used to describe other codes
b) Hence, it wouldn't can be linked to quotations.
c) It would be a variable code. Options:
- 0/1 (yes/no, active/inactive, ...)
- integer (e.g. 23%, 8(group of eight people, set size = eight))
- string from a list (e.g. colors: red, blue, ...)
d) It would (can) be automatically applied to all children of a member
of a hierarchical ISA Tree. (Atlas.ti would create and link such nodes
to the members)
e) Dependent on the kind of variable, it could have different values
for different members
f) We would have a button for auto-hiding/displaying attribute codes.
This would help to reduce the visual complexity of our network, when
the attributes aren't relevant in the respective situation.
g) ...

Second Way: Attributes Frames (See picture)

When we currently double click on a code node, then we can read
its comments.

a) Imagine a similar kind of activation would open/close a little table.
b) The table would contain two main columns: Attribute Name and
Attribute Value
c) We could introduce as many attributes as we need.
d) Attributes would have a type: string or integer
e) We would define attributes globally and use them when we need.
f) Attributes would be a new kind of entity in Atlas.ti. They would also
have an own manager.
g) If a code is linked as child of a more general one using ISA,
then attributes would can be partially inherited. But the child could
also have other attributes and also different values.

These aren't elaborated thoughts... there would be something to do,
but I think this concept could enormously increase our possibilities
to define and represent things. Also imagine how many possibilities
of visualization in tables or other systems this would open...

Thanks and Best Regards