P3D Searching the Archives (long)

[Bob Wier <wier@xxxxxxxxxxxxxxxxx>]

|     Hi all,
|     
|     Due to problems with our SMTP server and an 
|     involuntary u n s u b s c r i b e, I'm missing
|     a couple of digests : 
|     PHOTO-3D #2381-#2386 and #2400-#2404
|     SELL-3D #284-#287 and #298-#300.
|     Is there anybody willing to send me a copy ?
|     
|     Many thanks,
|     
|     Peter van Zuijlekom
|     GouDDDa
|     The Netherlands

Hi Peter (and others). This is probably a good time to let the 
list know that I've reorganized the archives a bit. In response
to a suggestion (Hi Michael!) I've broken down the digest section
into sub sections (so that if you access the digest archives via
ftp via a modem connection it doesn't take forever to scroll thru
the list of files). Currently these are in 1,000 digest chunks
(ie, 0-999, 1000-1999, 2000-2999) (I'm thinking about going down
to maybe 500 digest chunks instead). This would make scrolling thru
the file list much faster. 

There are several new features of the GLIMPSE search engine which
make searching easier to use. In the note below a "hit" is a match
on a keyword you may be searching for in the archives.

1) The search panel is hierarchical - if you search at the top
level for a set of keywords (see below) then all of the lower directories
will be searched as well. If you'd like JUST to search the lower
directory, click on the link and use the search panel there. Searches
will speed up as you search more restricted directories.

2) There are now options for "near" or "far" searches when you use a
AND type keyword search. The "near" option asks for matches which occur on
the same line in the file, whereas the "far" option searches the entire
file and succeeds if the keywords match anywhere (even on different lines).
NOTE: the number of "hits" expands rather dramatically when you specify
the "far" option.

3) There are two additional links at the bottom of each search panel -
The first takes you to a list of the directory contents, which you can then
browse thru using your netscape (or whatever). This is useful if you
would like to look at a sequential series of digests rather than 
searching for keyword matches. Currently this link takes you to the
top level of the photo-3d archives, so you can move up and down 
the directory structure by clicking on the folders containing the
items you may need.

4) The second additional link invokes your ftp program to download a
specific file - this can be either via netscape (or whatever) itself
or an ftp client if you have one specified in your preferences/options/
helpers settings.

5) The search panel uses a modified "Regular Expression" match criteria.
Mostly what you will want to use is a semicolon to specify and AND
operation and a comma to specify an OR operation. 

6) You can also specify if you'd like to look for a word which must
be in a particular case (ie, Realist vs realist). The default setting
is case INsensitive.

7) You can also specify if you'd like to search using a "fuzzy" spelling
match, with various degrees of mismatch. NOTE: allowing "fuzzy" spelling
options dramatically increase search times as the search engine has
to try all possible combinations of spellings on a given word. The
default is to look for an exact match on spelling.

Typical search examples:

Suppose you want to see what people might have
had to say about using Kodachrome. You specify Kodachrome by itself
and get a number of matches. Click on any of the matches and the file
will be displayed on your browser with the keyword Kodachrome highlighted.
By default all occurrences of "Kodachrome" and "kodachrome" will match
(case insensitive). If you only want "Kodachrome" (tradename) then
specify case sensitive matching.

Suppose you'd like to see all occurrences of people talking about
using Kodachrome with Realist cameras. You could try specifying

          kodachrome;realist

with "case INsensitive" and "near" searching. All occurrences of these
two words on the same line will be displayed. If you specify "far"
matching, all occurrences in a given digest will be displayed even
if they occur at different places in the file.

Suppose you'd like to see any discussions of the Besplaca? Belplaca?
camera. Specify case INsensitive and allow fuzzy matching with, say,
2 misspellings.

To answer Peter's request, you could move down to the subdirectory
for digests 2000-2999, and specify

      digest;2381

which would find that particular digest. Or you could switch over to
the sequential file access via one of the links at the bottom of the
search panel, and retrieve the digests that way. The second way is
more efficient as there are a LOT of matches occurring on the "digest"
keyword which then must be disqualified because "2381" is not present,
which take a lot of time via a keyword search.

Notes on Efficiency of searches:

Generally, the search engine will not search on "plain" numbers,
so specifying just 2381 will not work. The reason for this is that 
allowing plain number searches result most times in an overwhelming
number of hits being generated from header lines in mail messages,
and so forth. 

The more AND (semicolon) conditions you specify, the slower the
search.

The more misspellings allowed, the dramatically slower the search.

Specifying FAR matching slows the search (you may wish to try a 
NEAR search first, and if that is not satisfactory, then try a 
FAR search).

There are default limits as to how many matches are displayed
within each file, and how many total files may be matched. These
limits can be increased, but if you are on a (slow) modem connection,
it will take awhile to see the results of the search if you get
a lot of "hits".

If you specify a wide search criteria, the server may appear to 
"hang" and not be responding. If you get impatient, just hit stop
on your browser to cancel the search and narrow the criteria down
a bit.

The archives are available at

http://calcite.rocky.edu/photo-3d

(Note - for the time being we don't have a "www" in front of the
machine name - it works fine without it).

Questions as always to 

mailto:3d-moderators@xxxxxxxxxxxxxxxxx

or 

mailto:wier@xxxxxxxxxxxxxxxxx

THANKS

               Bob Wier
         mailto:wier@xxxxxxxxxxxxxxxxx
  09:32 AM  Tuesday, November 18, 1997
     Rocky Mountain College, Billings MT.
 keeper of the Photo-3d and Overland-Trails
mailing lists and the USA GPS Waypoint server 



------------------------------