On 3/13/2021 1:49 PM, Chuck Guzis via cctalk wrote:
> I'm sure that Fred could tell you about doing serious research.
On Sat, 13
Mar 2021, ben via cctalk wrote:
I suspect that means spending money, or travelling
around.
Well, I don't know from "SERIOUS".
and, I don't spend money, and hardly never travel, anymore.
Never pass up the chance to hear from those who were THERE!
Folk such as Chuck are the greatest information resource there is.
You might not get much response, but try to contact those who were
involved, or at least present.
And, in the spirit of teaching to such eggs, . . .
In using information retrieval search engines,
PRECISION VS RECALL are at the expense of each other. Go for maximum
RECALL, because you can always do your own relevance ranking to increase
PRECISION of the recalled set, but only if you got the maximum RECALL to
have the hits in your set.
(My biggest contribution was pointing out that when quantifying the "best
possible" results, one needs to also quantify the "worst possible".
Nobody agreed that the integral under the curve was a good quantification
of a RECALL/PRECISION curve.)
And, always remember that because search algorithms are relevance ranking
of overlaps with search terms, synonyms, or co-appearing words, therefore,
instead of putting a "SUBJECT" in the query, you need to put in whatever
words seem most likely to be in the result that you are looking for.
Even if that means getting some Apatasaurus when you were looking for
"dinosaur" mainframes.
--
Grumpy Ol' Fred cisin at
xenosoft.com