RE: docster: instant document delivery

["Dovey J, Mr" <pjcd@maties.sun.ac.za>]

Hi,

I know I am replying to this email a little late, but that is just how my life is running at the moment (a little late <g>)

 

I too have read Daniel's article on Docster, and I have been following with great interest the various stories about Napster, GNUtella and FreeNet online. I have a running GNUtella client and have installed the freenet experimental version on a Linux box. Both are interesting for various reasons although there are numerous suggestions for improvements to the protocol etc (GNUtella NG et al)

 

I have also been experiencing some of the limitiation inherent in the Ariel system, not least of which is the cost of purchase and the closed model for extension.

 

Being at heart a hacker (in the original sense of the word) I have started working on something in which I want to try to combine the best features of all of these systems.

From Ariel, I like the idea of the ability to REQUEST a document and then to have it SUPPLIED (ie Scanned etc) and from Prospero I like the ability to deliver the requested document back to the patron (on a server with a notification via email)

 

From GNUTella I like the idea of automatic discovery of other servers on the network and the manner in which messages are passed around this network with the ability for each client to process the messages and decide how to handle them within the bounds of the protocol. What I don't like is the requirement that the client has to connect directly to the source to retrieve the document nor the limited searching capabilities  and the inability to describe the files that are added to the network.

 

From FreeNet I REALLY like the idea that documents that are requested over the network are dumped at each node through which they pass which has the effect that documents rend to "clump" or be cached at servers near to the users that are requesting them.

 

With all these I ideas, I have started working on a program which I call "DoveDoc" in which I hope to onclude these features and a few more. My initial "vision" is that this could be used only by the various branch libraries within the institutions in our Consortia for document sharing, but seeing as I have a preference for the OSS model, of course I will make what I develop available. If this develops into something that is usefull generally, then so be it <grin> 

 

I would be interested to hear any views that anyone might have on this.

 

I am working in Visual Basic and I am at the very beginning of the development, but if there is anyone interested in seeing the status so far and/or working with me on this, then I would be only too happy to hear from you.  

 

John

 

---

John Dovey

Assistant Director (IT)

Library Services

University of Stellenbosch, South Africa

-----Original Message-----
From: Frank Walker [mailto:walker@nlm.nih.gov]
Sent: Tuesday, May 23, 2000 2:41 PM
To: pjcd@maties.sun.ac.za
Subject: Re: docster: instant document delivery

I've just read Daniel's paper on docster and find it to be very interesting.

From a technical standpoint, there should be few problems in getting such

a system up and running.  However, there are a few details that would need

to be worked out.  For one, there is the problem of identifying journal articles

that users receive.  In my world, which is biomedical, journal articles are

identified through the MEDLINE unique identifier.  This is assigned to each

article in any of up to 4,000 biomedical journals.  Docster should work well

for a group of journals like this whose articles are uniquely identified.  However,

what about the remaining 70,000 biomedical publications whose articles are

not indexed?  Is it possible to create a method of uniquely identifying articles

that people receive that are not routinely indexed such as those in MEDLINE?

For that matter, what about the non-biomedical world; is there any unique

identifier for journal articles in other fields?

 

Another potential problem with docster is how long will users keep copies of

their electronic material on their computers (in this case, journal articles)? 

In my experience with DocView software, most users tend to print their

articles after they are received; after that they may or may not delete the

electronic copy from their computer.  I suppose that if users see the benefits

of docster, they will try to keep electronic journal copies for longer periods of time. 

Eventually, hard disks become cluttered with bitmapped images, and users will

want to remove older articles.  The oldest articles will tend to become deleted first

from user's computers, which is probably okay, since researchers tend to

request recent publications from their libraries. A better model for document

retention might be Carl Uncover' central storage scenario, in which all articles are

kept forever, and the collection keeps on building without end.  In that case, old

articles will never be lost.

 

Another minor stumbling block is the method by which documents are received and

the disk directory where they are stored. Today there are three: MIME email attachments,

Ariel FTP and the WWW.  While the first two methods have InBoxes, where received

documents are stored until deleted, there is no InBox for web deliveries.  Documents

received through the web may be either printed directly via a client such as Adobe

Acrobat Reader, or stored first on disk.  But if it is stored on disk, there is no consistent

directory for storage.  Users of docster would have to configure their docster client software to

point to the directory (or directories) where documents are stored.  Would docster be

intelligent enough to be able to extract an attachment from a MIME email message?

Or would users have to do this themselves and save all document deliveries into a

special directory that docster would use?  In my experience, I've found some users who don't

have the slightest idea how to save a MIME email attachment.  This is all food for thought,

and details would have to be worked out.

 

Speaking of Carl Uncover, what happened to the lawsuit?  I saw that Carl Uncover was sold

to another company earlier this year, but did it affect the lawsuit?  Last I heard, Carl was

found guilty and was going to appeal.  It looked pretty bleak for a company that had a great

idea.  Apparently it was not enough to get permission from 6,000 publishers to keep electronic

copies of journal articles from their publications...it was also necessary to get permission from

all article authors, who might be dead or alive, and whose numbers might total in the millions.

All this has implications for docster, and I am sure that without a legal team from day one, docster

might suffer the same fate as napster and Carl Uncover, and be smothered with lawsuits.

 

Frank Walker