SIGIA-L Mail Archives: Re: SIGIA-L: Measuring IA
Re: SIGIA-L: Measuring IA
From: James Weinheimer (jamesw_at_princeton.edu)
Date: Wed May 17 2000 - 08:55:31 EDT
Louis Rosenfeld wrote:
> Alternatively, we can measure user satisfaction. But this is not the same
> thing as measuring whether or not users find what they are looking for.
> knowing any better, a user might express satisfaction with the inaccurate
> incomplete information that a system provides. Ignorance is bliss, but
> blissful users don't necessarily indicate a successful information
This is the key point when using a marketing goal (user satisfaction) for
purposes where it just doesn't apply.
Here is a real-word example in present bibliography. Someone may want items
World War II, so they search for the keywords:
World War II.
They will get a lot of records. (In Princeton's catalog, 2417 hits). This is
more than they can handle.
Now the question is: Is the searcher happy?
Perhaps, but that doesn't end the argument. If the searcher made the correct
search in the standard bibliographies:
World War, 1939-1945
s/he would get many, many more hits. (In Princeton's catalog, tens of
The question changes to: Is the searcher still happy with the first search?
One way of looking at this is: if searchers are to remain happy with the
search, then they cannot know about the second possibility.
Do people really understand that when they do a search on, e.g. World War
they are only searching for *three random words* instead of searching the
*concept* of WWII? Fuzzy searching can help, but still has a lot of holes.
So, I think Lou is right on target. User satisfaction is nice, but in the
is practically irrelevant for the measurement of IA.
This archive was generated by hypermail 2.1.2
: Sun Nov 23 2003 - 22:54:19 EST