Comparison between Gopher, WAIS and WWW
Hank Nussbacher
Gopher/WAIS/WWW and to see which would be the most appropriate to
use. I would like to present my findings so that others may benefit.
Gopher is a menu driven system. Each Gopher server has links to
other Gopher servers so that one can "seamlessly" bounce from one
server to another with no perceived change of environment. Gopher's
down-side that I have determined is in finding the proper resource.
There now exists dozens of Gophers, each with dozens of sub-menus.
Somewhere buried in some Gopher server on the other side of the
world might be some very important resource that you have been
looking for. Unfortunately, there is no easy way to know about that
resource unless you start a lengthy and time consuming Gopher tour.
WAIS is a method of searching indexed text. Someone can come along
and take a hundred files on the topic of numerical analysis and
index them into one .SRC file. New resources are announced in a
special indexed database called "directory-of-servers" which solves
the Gopher problem mentioned above.
WWW is hypertext. It goes beyond Gopher which can display menus and
beyond WAIS by allowing pointers and chains within the text.
Hypertext links can lead you down a path that may be fruitful or it
may be a dead end. Anyone who has used Hypertext on a multimedia
system knows the power that it affords the user.
So which is for you?
I view Gopher as the SMTP/SNMP of information retrieval. I would
call Gopher - Simple Information Retrieval (SIR). That is the
reason it is far more popular that WAIS and WWW combined. But why
is that so? Let us take an example. You are mandated to create an
information system for your university. You would like to include,
among many things, local bus line timetables, restaurants in the
area, and apartment rentals.
We could set all the data into a WAIS index and the user could
search of the word "McDoogal Street" and find which buses run there,
which restaurants are there as well as apartments to be rented. But
if the restaurant is located on a side street off of McDoogal Street
it will not come up in the search. How is a student to know where
is a good place to live near the campus? What words would he/she use
to search with? Clearly WAIS is not the solution.
Lets try WWW. We could create the following Hypertext: I could find
an apartment to rent on McDoogal and then click to see the closest
bus line and then click its time schedule and then click to see what
Chinese restaurants are nearby. The possibilities are endless. It
is super-cool. But practical - no way. No university would invest
the manpower to hypertext all this info (which takes enormous time
and needs to be constantly updated) and add in all the necessary
textual links. When is Hypertext needed? Encyclopedias and
reference books are ideal to be hypertexted. You do it once and you
don't have to update it the next semester. You can even buy data that
has already been hypertexted - as long as there is enough of a
demand from a commercial source. But to sit and hypertext the QX-1
bus line is an extreme waste of resources.
This is where Gopher comes in. A user is given a menu with 3
submenus: Bus lines, restaurants, apartment rentals. The user goes
to the document, brings it up, scans it, perhaps prints it to show
to a friend. A secretary can be in charge of updating the file. It
is a straight ASCII file with no special codes or pointers. In
other words, it is easy to maintain and easy to retrieve.
The recent excellent Krol book "The Whole Internet Users Guide and
Catalog" comes out in favor of WWW over Gopher. I suspect this is a
common trap us techo-dweebs fall into. The more features, the more
powerful, the better the system has to be. It is the same pitfall
people fell into 10 years ago when they ran out to buy a PC to
balance their checkbooks.
If you wish to set up a campus information system that is simple to
use, and simple to maintain, I suggest you look into Gopher. If you
wish to support a certain campus faculty (like medicine or law) that
has many online documents that they wish to be able to search
easily, look into WAIS. If you have reference books and extra
manpower available look into WWW.
Hope this helps.
Hank Nussbacher
Bar-Ilan University
Israel
ha...@vm.biu.ac.il
Barry Margolin
>We could set all the data into a WAIS index and the user could
>search of the word "McDoogal Street" and find which buses run there,
>which restaurants are there as well as apartments to be rented. But
>if the restaurant is located on a side street off of McDoogal Street
>it will not come up in the search. How is a student to know where
>is a good place to live near the campus? What words would he/she use
>to search with? Clearly WAIS is not the solution.
I agree. WAIS was designed for searching large text databases; it derived
from our work done for Dow Jones News/Retrieval on searching their news,
magazine, and journal library. It was intended to be a replacement for
error-prone boolean searches and keyword lookups, and relevance feedback is
an important aspect of this. It doesn't work well, in my opinion, for
structured databases with terse "documents". I can't imagine using WAIS
for a geographic database as in your example.
>Lets try WWW. ... You can even buy data that
>has already been hypertexted - as long as there is enough of a
>demand from a commercial source. But to sit and hypertext the QX-1
>bus line is an extreme waste of resources.
I think the hope is that as such databases become more common, more
databases will be available from 3rd parties, rather than requiring the
database maintainer to hypertext it themselves. Thus, the local transit
authority would hypertext the bus schedule as a service to the community.
If they do things right, they could use the same hypertext database to
generate the printed maps and schedules that you pick up at stations.
>This is where Gopher comes in. A user is given a menu with 3
>submenus: Bus lines, restaurants, apartment rentals. The user goes
>to the document, brings it up, scans it, perhaps prints it to show
>to a friend. A secretary can be in charge of updating the file. It
>is a straight ASCII file with no special codes or pointers. In
>other words, it is easy to maintain and easy to retrieve.
I don't see how this solves the "side street" problem you mentioned in the
WAIS paragraph quoted above. If the restaurant list is sorted by street
names, you're not going to find restaurants on side streets.
Some related issues are raised in the latest CACM (I don't recall the
month, but the cover article is about expert system shells), in the
Practical Programmer article about "Hat Racks of Data".
--
Barry Margolin
System Manager, Thinking Machines Corp.
bar...@think.com {uunet,harvard}!think!barmar
Tim Berners-Lee
<HA...@BARILVM.BITNET> writes:
> Gopher/WAIS/WWW and to see which would be the most appropriate to
> use. I would like to present my findings so that others may benefit.
He then goes on to make some good points but to make a
FUNDAMENTAL MISUNDERSTANDING that I would
like to correct now. He summarises the situation by saying:
> Gopher is a menu driven system.
..
> WAIS is a method of searching indexed text.
..
> WWW is hypertext.
Let's be more accurate:
WAIS allows searches.
Gopher allows searches AND menus.
WWW allows seraches AND menus AND hypertext.
There are lots of Gopher servers which have text indexes to them.
The gopher protocol allows a search to be made. But WAIS does not allow
menus. OK?
Same with WWW. WWW has menus -- to the reader they are just simple hypertext.
And WWW has searches. But WAIS and Gopher don't have hypertext.
The big misunderstanding is to conclude:
> Lets try WWW. We could create the following Hypertext: I could find
> an apartment to rent on McDoogal and then click to see the closest
> bus line and then click its time schedule and then click to see what
> Chinese restaurants are nearby. The possibilities are endless. It
> is super-cool. But practical - no way.
But you *Don't Have To!* Sure, if you *want* to invest the time, do!
But you just as well put up a menu system using W3. As an information
provider you have the choice. You can just use directories as menus
as in Gopher. You may find that for the first few pages you prefer to
customise as hypertext because that page will be read by moer people
than anything else. But you can be lazy and just put up a whiole
directory tree. If you browse around th web and think that all the
hypertext has been hand-edited you are getting very much th wrong idea.
Noone has the time. Most of the data out there is generated by some
server which makes it LOOK like hypertext when in fact its just text files
or news or relational databases or Gnu Info or or or ....
> No university would invest
> the manpower to hypertext all this info (which takes enormous time
> and needs to be constantly updated) and add in all the necessary
> textual links. When is Hypertext needed?
Wrong wrong wrong. Hypertext turns out to be a neat idea BECAUSE
it can represent menus to start with. When you really get into it, you often
find that you need a little more. (Look at the Panda project which started
just so as to get a little descriptive text above a gopher menu!).
> This is where Gopher comes in. A user is given a menu with 3
> submenus: Bus lines, restaurants, apartment rentals. The user goes
> to the document, brings it up, scans it, perhaps prints it to show
> to a friend. A secretary can be in charge of updating the file. It
> is a straight ASCII file with no special codes or pointers. In
> other words, it is easy to maintain and easy to retrieve.
>
> The recent excellent Krol book "The Whole Internet Users Guide and
> Catalog" comes out in favor of WWW over Gopher. I suspect this is a
> common trap us techo-dweebs fall into. The more features, the more
> powerful, the better the system has to be. It is the same pitfall
> people fell into 10 years ago when they ran out to buy a PC to
> balance their checkbooks.
The guy who bought the PC to balance his checkbook didn't realise
how the power and flexibility of the PC could help him *until he
brought it home*.
> If you wish to set up a campus information system that is simple to
> use, and simple to maintain, I suggest you look into Gopher. If you
> wish to support a certain campus faculty (like medicine or law) that
> has many online documents that they wish to be able to search
> easily, look into WAIS. If you have reference books and extra
> manpower available look into WWW.
Run Gopher server becuse of the excellent work which has been
put into making them easy to run on a Mac for example.
WAIS-index your data when you have a moment and a unix box.
A W3 client can access all these, and your FTP archive too.
Install W3 clients (like Viola), and you will have the option of putting links
or text formatting in when you want to offer a better service. Later, you may
want to write some hypertext to show people around who don't know what the
thing are looking for is called. Or you may want to write a shell script to put
you favorite database on-line. Or at least as in introductory page to explain
what your institute does. Then you can point them into the Gopher tree... and
W3 readers will follow the links without realising the difference. So a W3
server will be appreaciated by your readers.
> Hope this helps.
I hope I have explained that one OK. I seem to have to do it from time to
time... The comparison of the 3 systems is a W3 FAQ by the way (on the web).
Tim
> Hank Nussbacher
> Bar-Ilan University
> Israel
> ha...@vm.biu.ac.il
_____________________________________________________________
Tim Berners-Lee ti...@info.cern.ch
World Wide Web initiative (NeXTMail is ok)
CERN Tel: +41(22)767 3755
1211 Geneva 23, Switzerland Fax: +41(22)767 7155
я