Re: Text Searching in Oracle? - MIME translated

From: Eric D Pierce <sac50216_at_saclink1.csus.edu>
Date: 1995/04/12
Message-ID: <3mhh29$1qh_at_news.csus.edu>


MIME translated version of this message reposted (my Unix reader choked on MIME, so I had to send this to a MSDOS PC email program for translation, I hope repost this prevents others from having a similar problem):

From: panderso_at_ottawa.net (Paul Anderson) Newsgroups: comp.databases.oracle
Subject: Re: Text Searching in Oracle?
Date: 12 Apr 1995 04:00:53 GMT
Organization: Internet Access Inc. <613> 225-5595 Lines: 121
Mime-Version: 1.0
Content-Type: multipart/mixed;

     Boundary="*-*-*- Next Section -*-*-*" X-Newsreader: WinVN 0.92.1

--*-*-*- Next Section -*-*-*
Content-Type: text/plain

In article <3m1b86$pre_at_news.csus.edu>, sac50216_at_saclink1.csus.edu (Eric D Pierce) says:
>
>Michael G. Barbitta (michaelx_at_leland.stanford.edu) wrote:
>: In article <3luvt9$q1d_at_ttis.thomtech.com> Dave H <dheissner_at_thomtech.com>
 writes:
>: >From: Dave H <dheissner_at_thomtech.com>
>: >Subject: Text Searching in Oracle?
>: >Date: 5 Apr 1995 20:51:53 GMT
 

>: >Are there any third-party products available which provide text search
>: >capabilities for Oracle data? Has anyone does this and what would be
>: >a good approach to take?
 

>: >TIA
... text deleted ...

>
>I hope an expert or someone with experience using these tools
>will speak up with a clarification.
>
>I'm pretty sure that there are other 3rd party products out there
>too. I have seen messages on the discussion groups related to
>image data being stored in fields defined with "raw" data types?
>
>Sorry for the vagueness.
>
>Any feedback is appreciated,
>
>EP
>Database/Network Technician
>California State University, Sacramento
>
> sac50216_at_saclink.csus.edu
> or
> PierceED_at_csus.edu
>
>cc: dbmoore_at_us.oracle.com (CDE2 Marketing God)

--*-*-*- Next Section -*-*-*
Content-Type: Text/Plain

Oracle has had a full text retrieval capability for about 4 years now. Until recently it's been called SQL*TextRetrieval. The product is basically an API that allows a SQL*Forms application or a 'C' program to build SQL queries that search through text in an intelligent manner. SQL*TextRetrieval is built on top of an Oracle database. Text data may be held in Long columns or CHAR columns. Also, because of the limitations of Oracle Version 6 long columns, extended text fields were created which allowed a piece of text to be as long as a user wanted (i.e. not limited to 64k).

Within the past year, the product was renamed TextServer3. This is more than a rename. It has been rebuilt almost completely to take advantage of client/server features in an Oracle database. As it stands now, a client (MS-Windows or Motif) requests text services (such as search, create, index, update) from a Text Server (currently SUN/OS, soon to be Solaris, HP, RS/6000, MVS, perhaps others). The
Text Server is responsible for executing stored procedures in an Oracle7 database and retrieving
information (text or structured data) based on full text searches. Assuming you have an EMP table
with a long column called RESUME that contains resumes, then an example of a text search is:

    SELECT sal, empno, name, comm, resume from

       emp where
        sal > 3000
      and
      resume contains
                 'cobol' and 'manage*' and 'Oracle'(1,0)'expert'

This query asks to return data based on the fact that the resume contains the word COBOL, words starting with "manage" and "oracle' within one word of "expert". Also I've incorporated a non-text argument into the where clause just to round it out.

For the most part this SQL statement would be created by the Forms app and the end user wouldn't have to see any of this.

It's important to remember that TextRetrieval is part of the development environment and any ugliness is probably at the Forms design level. SQL*TextRetrieval can be anything a developer wants.

In addition to full thin client/fat server capability, TextServer 3 incorporated a product called CONTEXT which we purchased from an outside source. Among other things, CONTEXT allows us to build themes into documents, so users are no longer looking for words, but meanings. This is extremely powerful. Beyond building themes it also allows us to build precis' of documents and, in fact, we will reduce a document down 5 different levels, as a user requires. ( In fact, Context has 16 different reduction levels, but the TextServer developers decided 5 was sufficient for the TextServer product.

This is a really brief outline of the TextServer product. If you'd like more information, please feel free to contact me at panderso_at_ca.oracle.com or panderso_at_ottawa.net or talk to your sales rep.

--*-*-*- Next Section -*-*-*-- Received on Wed Apr 12 1995 - 00:00:00 CEST

Original text of this message