I apologize for the sketchy nature of this proposal - this is all I have
time for tonight.
Abstract
--------
Volunteers are wanted to help building a comprehensive, powerful, easy to
use and innovative documentation tool for LISTSERV based on gopher and
the LISTSERV database functions. Only moderate knowledge of LISTSERV is
required for the bulk of the work ("task 1" below). Contributions are
expected from list owners and user support people who could eventually
recover their investment in time through a reduction in end user
questions.
Description
-----------
Most of the questions asked by regular end users, list owners and
LISTSERV maintainers alike have been answered many times on mailing lists
such as LSTSRV-L and LSTOWN-L, or are available from release notes and
other documents that can be easily searched using the LISTSERV database
functions. Unfortunately, users are too lazy to learn how to use the
database functions and keep posting the same questions to the lists (or,
in more optimistic terms, one could say the initial investment of
learning how to use the database may not seem warranted to a user who is
confident he will not need to ask more questions in the near future). In
addition, because the same questions are asked all the time, database
searches will typically find dozens of answers, some of which are
incorrect or incomplete. The users want "safe", "official" answers.
FAQ files would probably not help, due to the volume of questions. There
would have to be thousands of lines of FAQ, probably split into several
files, and users who are not familiar with editors will have a hard time
finding their question.
The project is to have a number of independent groups extract answers
from the list archives and consolidate them into comprehensive, up to
date and accurate "answer sheets" which will then be made available via
gopher in the usual tree-based fashion. The "credo" of the project is
that, if this tree structure is made intuitive enough, users will easily
find the information they are looking for in little more time than it
would take them to send mail to user support or to a mailing list - and
they may find answers to other things they were curious about on the way
down the tree.
If this proves successful, the project can then be extended by having
actual tutorial items inserted in the hierarchy.
The organization
----------------
In order to get the work done faster and to allow people with a moderate
amount of time to contribute, the work is carried out in parallel by a
number of independent groups. Each group is coordinated by a
knowledgeable LISTSERV user, preferrably with user support experience.
Ideally there should be another 5-10 editors in the group, preferrably at
the same geographical location.
Task 1
------
The group coordinator is sent 10-25,000 lines of source material, as per
his request. This will normally be list archives in chronological order.
The coordinator splits the material in chronologically contiguous files
of a manageable size and spreads the work among the editors for "task 1".
The editors simply edit the files and identify question-and-answer
sequences. The original question is extracted from the posting in which
it first appeared, with the name of the poster removed and '>' signs
placed in column 1. The 'Date:' and 'Subject:' fields of that original
posting are kept, and the editor changes the subject if it is not
meaningful. All other header fields or signature data are removed. The
editor then looks for answers to the question, discarding duplicates and
answers which later turned out to be incorrect. One or several answers
are selected and appended to the question file; the 'From:' line
containing the name of the person who answered is kept, all other header
fields are removed and so is unrelated information. Each question is
filed into a separate disk file.
When the editor is done, he reviews his list of questions and removes or
merges duplicates. Except for "official" announcements from release notes
or the like, the answer sheets should not exceed 2-3 pages. When
everything is ready, the editor forwards all the files to his
coordinator.
As you can see this is something which can easily be asked of students
working for user support or similar. Three sessions of a couple hours
each should be able to accomplish quite a lot without boring the students
too much - this isn't worse than reading a backlog from some large usenet
groups to find the answer to some particular problem. The students can be
motivated by being told that they will learn a lot about LISTSERV in the
process (and LISTSERV knowledge is usually a plus if you are looking for
a part time user support involvement to help paying your studies). And,
of course, everyone will receive credit.
Task 2
------
The coordinator receives all the question files from his group and first
repeats the process of eliminating duplicates. An EXEC will be provided
to make a subject list from the question files. Questions which have
become obsolete are eliminated or replaced with a short writeup if the
coordinator feels the question is likely to be asked anyway. After
checking the validity of the answers, the 'From:' fields are removed and
the various answers are merged into a more coherent writeup, if
necessary. The 'From:' fields are only kept when it is important to know
who provided the answer - for instance in comparisons between LISTEARN
and LISTSERV it will be important to identify replies from Turgut and
from me.
Above all, the coordinator makes sure the subjects are meaningful and
makes a first rough classification of the questions according to a short
document to be provided before the project starts. Minimally, the
questions will be marked as maintainer, list-owner, general user, mailing
list, file server, database, etc. A few XEDIT macros will be provided to
make this task easier. The coordinator finally ships the couple hundred
remaining files to the central coordinator.
Task 3
------
The central coordinator(s) repeat the process of eliminating duplicate
questions and are responsible for quality control and classification of
the questions in a gopher tree. There is a fourth task to actually
implement the whole thing on a gopher server and make a LISTSERV database
interface, but it is not described here.
How to apply?
-------------
Reply to me and let me know which sort of tasks you would be willing to
do and how much volume you think you can handle. You can also reply to
the LSTSRV-L list if you want to make comments about the proposal. Do not
reply to LSTOWN-L unless your comments are specifically aimed at list
owners.
Eric
|