The Computer Science Bibliography server as motivational example for DASL.

The Computer Science Bibliography server is a great example of an internet
search system of the kind that DASL ought to support.  

It's at http://liinwww.ira.uka.de/bibliography/

The CSB is a collection of nearly 1 millons bibliographic references to
computer science, some online, others only in paper.  It has a Web user
interface offering both a 'simple' and an 'advanced' user interface.  The
two UIs are built using Glimpse and FreewaisSF, respectively.  They are
good examples of the power of current freely distribuyed text search
engines, and the CSB as a whole provides some ideas about what features are
and are not useful in large text search systems.

There are at least two ways one the CSB could use DASL.  First, if CSB
supported DASL, one could experiment with new search user interfaces
independent of the underlying search engine.  Second, the CSB collects its
information from a number of sources on the Internet.  If those sources
supported DASL, it might make it easier to gather the information.  If CSB
supported DASL, one could build on top of it (to construct a larger unified
database.)

So it makes sense to look at CSB and see what features it has and which of
them could be done using DASL.  Of course, just because CSB has a feature
does not mean that users actually use it.  Some of them may not be so
important.

At the most basic level, DASL seems be good match for CSB.  The various CSB
fields (author, title) could map to DAV properties.  While I did not find
any evidence of full text indexing while browsing I suspect that there must
be at least some abstracts indexed somewhere.  

Query options include:
  choice of case-sensitive or insensitive matching.  
  limit of number of results. 
  partial match. 
  comparison operators (<, =, >) on fields (in Advanced interface only)
  choice of stemming (advanced interface only).  

DASL can support all of these except choice of stemming.

The CSB appears to go to considerable trouble to support characters not
found in USASCII, e.g. diacritics, accents, and letters from German.  For
the most part, the designer has chosen to use the LaTex conventions for
these characters, thus one searches for H{\"o}pfner.  I would imagine that
the Unicode support in WebDAV would be more convenient, but perhaps users
have gotten used to the LaTex conventions.

In sum, DASL seems reasonably close match to the needs of CSB. If anyone
has a different opinion it would be useful to hear of it.  Also if you know
of other large systems whose design should influence DASL, please mention
them in this list.

I have CC'd the designer of CSB, Alf-Christian Achilles, on this message.
If I have misunderstood his system perhaps he will correct me.

Jim


------------------------------------
http://www.parc.xerox.com/jdavis/
650-812-4301

Received on Wednesday, 18 November 1998 18:26:08 UTC