cni-directories: Re: Information Overload
Re: Information Overload
Edward Vielmetti (emv@MSEN.COM)
Tue, 11 Feb 1992 00:06:54 -0500
Date: Tue, 11 Feb 1992 00:06:54 -0500
From: Edward Vielmetti <emv@MSEN.COM>
Subject: Re: Information Overload
> Depends which one you use. Like any software It takes getting used to. I
> find LISTSERV retrieval language clumsy. Not that archives of usenet group
> are much better. Ftp the file back and use grep or awk to find the good
> stuff ... shudder! I put mine into hypercard to get full boolean.
The clear trend is to make all on-line archives of usenet groups,
mailing lists, and electronic serials full-text searchable with
an interactive protocol such as WAIS or mediated by something like
gopher. Part of the hard problem is getting the information that
a particular mailing list (or newsgroup or serial) is essential
to staying current in the field and has enough stored up quality
discussions to be useful as a reference, *and* finding it on-line.
Hard problems given how decentralized everything is. Projects
like CNI TopNode and the CICnet "network information resources"
committee help some, but clearly there's a lot of work needed on
all fronts before we're past the "ftp it back and grep" stage.
> There about 1500 usenet groups and (I think) about the same number of
> LISTSERV groups. You can search for either in LIST of LISTS or ACADLIST.
> Fortunately there is a WAIS server which gives keyword searching for
> either.
A serious problem with the WAIS server w/r/t Usenet groups is that
although the groups usually have a charter or other descriptive
information that goes with them, there's not such a thorough
effort that's been done to date to catalog them as well as e.g.
the ACADLIST (ARL mailing list collection) work does to group
them together by function. Too many lists of newsgroups are just
big long alphabetical efforts without more than 5 words of text
about each, certainly not enough directory informatino to find them.
>>The naming structure of news groups is difficult to understand and illogical.
>>At least when I subscribe to a listserver I can get a description of
>>it relatvely easily.
>One is classified and the other is not. A bit like call numbers really.
>It's all just marking and parking.
I would lay claim that an experienced cataloger, armed with a big
pile of disk space, some good full text searching tools, and a native
guide to the internet, could come up with reasonable Dewey numbers for
most Usenet newsgroups in something less than a few hours apiece.
At an hour a year that's one person-year full time plus support
staff to work the whole list. Double or treble the effort and you
can probably get a description of each one relatively easily.
The naming structure of the newsgroups has evolved over time.
It is no more illogical than the naming of BITNET nodes or the
Library of Congress call letters. There are a number of people
who have strong feelings about what newsgroups should be named,
and part of the creation process for a newsgroup is an open discussion
of what it should be called. (see news.groups or news.announce.newgroups
for details). This is an open, raucous, imperfect process with
not unreasonable results; enough people manage to find the right
group that they fill up a CD-ROM a month with their text....
It may well be that your local system has a rotten user interface
for netnews. It is hard to effectively present 2000 groups and
20 megabytes of text per day in a way that someone who is not
versant with the system can get their hands on. For some discussions,
LISTSERV is clearly appropriate technology.
--
Edward Vielmetti, vice president for research, MSEN Inc. emv@msen.com
MSEN Inc., 628 Brooks, Ann Arbor MI 48103 +1 313 741 1120
![[CNI Home Page]](/Images/home.gif)
|
