[SAC-FAST] Re: SAC Subcommittee on FAST

[qiangjin at uiuc.edu]

Hi everyone,

Below is Stpehen's message. 

Qiang

I don't want Arlene's list of questions for the sample review
to get lost in the discussion of chronological headings.  I
like the list as a way of measuring the fit between the FAST
headings and the item. We need to be careful for the second
question that we don't hold FAST responsible for questionable
LCSH heading-topical-subdivision assignments, though I doubt
there will be many of those. And for the last question
regarding chronology, I think we can evaluate the algorithm
for deriving periods from LCSH headings--specifically, are
there problems with significantly over- or understating the
chronological scope of the item using the algorithm. I agree
that this will not measure the usefulness of manually assigned
chronological headings, but that's not what we'll have to review.

The other question we'd like to explore with the sample is how
well the FAST headings work collectively for searching; but
that will be harder to do. Such an evaluation would have to
consider interface design, index configuration, sample size
and focus--lots of variables not directly related to FAST. The
other large question of what kinds of headings would be
assigned if we let FAST loose in a world longing for
unfettered simplicity can't be answered directly by the
sample; but maybe indirectly--we could consider what FAST
headings we might have chosen for an item, and why.

We also need to talk about how many records we'll be able to
review. Five hundred records each is sounding awfully
ambitious, unless we can come up with a tiered review--say,
100 records closely and quantifiably reviewed, 400 records
skimmed for anomalies and general impressions.

Stephen