SIGIA-L Mail Archives: SIGIA-L: RE:Natural language definitions
SIGIA-L: RE:Natural language definitions of requirements
Date: Thu Jun 29 2000 - 12:20:37 EDT
Natural language is a double-edged sword.
While it has the power to better capture subtleties and shades of
meaning, it is exactly these properties which tangle the requirements
If you show a textual use case to 5 different developers, the chances
are high that you'll get 5 different implementations of the "same thing".
>From the s/w engineering POV, the whole idea in using a formal notation
such as UML Use Case diagrams (there are, of course other ways!) is to
remove any ambiguity.
You have to start somewhere, which is why brainstorming and writing it
down is a great first step. However, this must be refined in order to get
an exact definition.
Related to my other comments in other posts about this general discussion,
IF the kinds of information we are trying to capture in IA activities
are others than those well-handled by UML or other formal notations, then:
1) What are they?
2) What makes them so darn elusive?
3) Are they "formalizable" at all? Are they more "artistic" things?
I offer "complexity" as a vague combo-answer to numbers 1 and 2, with the
attached comment that the phenomenon of complexity is definitely common to
both s/w engineering and IA. What the nature of "complexity" in IA terms is
I am still trying to fully grasp.
In the case of #3 perhaps there is really no answer and any discussion about
applying notations to try to formalize these things is hogwash and the
can end here.
From: Brian Sheppard [mailto:bboru_at_umich.edu]
Sent: Thursday, June 29, 2000 7:47 AM
Subject: RE: SIGIA-L: IA notation
Indeed, it may be that language and literature courses will prove to be
the most valuable 'tool' in this regard. Human language has the unique
capacity to express directly without forsaking complexity.
> However, in the meantime I have gone back to representative (as
> opposed to complete) written usage scenarios for capturing requirements;
> paper prototypes (of page layouts and the first 2-3 levels of structure)
> communicating designs.
> Andrew Gent
This archive was generated by hypermail 2.1.2
: Sun Nov 23 2003 - 22:54:20 EST