(0)RDBMS for free text indexing

From: Bjug Boyum <bjug.boyum_at_nsd.uib.no>
Date: 1998/07/15
Message-ID: <6ohkog$rfh$1_at_toralf.uib.no>#1/1

Hi,
We are considering several rdbms alternatives for a database with mixed structured and free-text data (an infosystem for documentation of research in Norwegian universities and colleges). Our legacy data resides in a dos dbms (Advanced Revelation). We have currently redesigned it to a relational schema, and have implemented it on ms-sqlserver (which is the rdbms we currently are using for other databases). The text fields are mostly research project descriptions, result summaries, publication references and the like. We are looking for a system with built-in text indexing capabilities, for making this kind of information searchable via the web and in our user applications. Most of the major RDBMSes provide text indexing today, but we haven't seen any good independent evaluation of the products regarding this kind of functionality. So, I would like to hear if anyone has opinions or real-life experiences with text indexing in the following systems:

Oracle 8.0 (with context cartridge)
Sybase ASE 11.5 (with Verity snap-in)
Informix 9.1x (verity, fulcrum datablade betas, excalibur text datablade) IBM db2 5.0
(others?)

I generally like oracle and informix for their OO features (particularily arrays/multivalued fields, which we have had in Arev for over a decade, and used extensibly in our applications). However, Informix 9.1 has no replication facilities (whitch we really need), and do not support ObjectIDs and references (which we really want). As for Oracle, their prices are a bit over the edge for us, and I also have an (irrational?) impression that oracle is more proprietary and difficult to administer than the others. I generally dislike the idea of running an oracle server in a relatively small-scale facility like ours. I really like sybase for its simplicity (and familiarity to ms-sql which i allready know), but have doubts about their 'snap-in' solutions. I'm also unaware of any plans they might have for implementing SQL3. The same goes for IBM, and for both sybase and IBM, the severe constraints on row size is also a major turn-off.

Comments anyone? The current online versions of our R&D databases can be seen at http://www.nsd.uib.no/english/research/ (warning: they suck!).

TIA,
    Bjug Boyum

---

Bjug Boyum
Norwegian social science data services
http://www.nsd.uib.no/english Received on Wed Jul 15 1998 - 00:00:00 CEST