[LWN Logo]
[Timeline]
From: mark dufour <m.dufour@student.tudelft.nl>
Date: Fri, 1 Dec 2000 07:55:39 +0100
To: letters@lwn.net
Subject: distributed open directory search engine for linux

Hello,

I would like to know if there might be a story in the following. Two other CS 
students and me have developed a new type of search engine for linux that we 
think might be revolutionary. My english is not that great so I'd rather not 
write an editorial myself. I would really appreciate a reply to this mail, 
even if it's as short as 'go away'. Thanks in advance.

Kascade is a novel type of search-engine, based on the Open directory 
principle. This means that anyone can structure a small part of the 
information on the internet, corresponding to their personal interests or
expertise. The resulting parts are placed in a large categorical structure 
that others can browse to search for information. Examples of Open Directory 
initiatives are Dmoz and Infomarker.
 
In contrast with other Open Directory initiatives, though, the parts that 
people maintain now reside on their own server. What results is a distributed 
(!) Open Directory. As with Gnutella, though, there are no fixed central 
servers. This implies that anyone can start a new structure. Actually any 
piece can be easily replaced, hopefully leading to competitive improvement. 
The distributed nature and lack of central control facilities make for a 
system immune to company or government control, much like Gnutella and the 
Internet itself.
 
Kascade directories are based on a specially developed file format, called 
DII. To browse Kascade directories consisting of DII files, a special browser 
has been created that runs only under Linux, called the Kascade client. Its 
visual size is as small as the average ICQ client. When a user has found what 
seems to be an interesting web page, the client fires up a web browser to 
view the page. It also has a built-in IRC client that enables users to chat 
with each other anywhere in a directory!
 
The DII file format features aspects comparable to 'components' and 
'functions' in programming languages, introducing 'abstraction' and thus 
structure 'reuse' into the Open Directory world. These components and 
functions themselves may even be distributed! Not less useful, so-called 
'queries' may be defined into parts, that query the same and/or other parts, 
and create new local hierarchical structures in real-time (as users browse a 
directory.) DII queries, functions and components can be used to easily 
create multiple search paths to the same information item.
 
Please visit the project site at http://kascade.org for much more 
information, screenshots and for downloading the Kascade browser.


- -- 
Mark Dufour <m.dufour@student.tudelft.nl>
Kascade <mark@kascade.org>