-------------------------------------------------------------------------

Oracle8 ConText Cartridge

Release 2.3 (Version 2.3.6)

RELEASE NOTES

December 1997

-------------------------------------------------------------------------

Copyright (C) 1997 Oracle Corporation

Primary Author: D. Yitzik Brenman

Contributors: Peter Bell, Chung-Ho Chen, Paul Dixon

This software/documentation contains proprietary information of Oracle

Corporation; it is provided under a license agreement containing

restrictions on use and disclosure and is also protected by copyright

law. Reverse engineering of the software is prohibited.

If this software/documentation is delivered to a U.S. Government Agency

of the Department of Defense, then it is delivered with Restricted Rights

and the following legend is applicable:

RESTRICTED RIGHTS LEGEND:

Use, duplication, or disclosure by the Government is subject to restrictions

as set forth in subparagraph (c)(1)(ii) of DFARS 252.227-7013, Rights in

Technical Data and Computer Software (October 1988).

If this software/documentation is delivered to a U.S. Government Agency

not within the Department of Defense, then it is delivered with

"Restricted Rights," as defined in FAR 52.227-14, Rights in Data -

General, including Alternate III (June 1987).

Oracle Corporation, 500 Oracle Parkway, Redwood City, CA 94065.

The information in this document is subject to change without notice.

If you find any problems in the documentation, please report them to us in

writing. Oracle Corporation does not warrant that this document is error free.

Oracle, ConText, SQL*Loader, SQL*Net, and SQL*Plus are registered trademarks

of Oracle Corporation. Oracle8, PL/SQL, and Gist are trademarks of

Oracle Corporation.

All trade names referenced are the service mark, trademark, or registered

trademark of the respective manufacturer.

----------------------------------------------------------------------------

**** NOTE ****

With each distribution, ConText provides a workbench consisting of tools and

utilties which enable:

-- easy, GUI administration of ConText from a client machine or a World Wide

Web-environment

-- powerful and flexible development of ConText applications for Windows NT,

Windows 95, Oracle Forms, or the World Wide Web

The ConText Workbench can be installed on any IBM-compatible personal computer

running Windows NT or Windows 95. The Web-based administration and

application development components can be used with either Oracle WebServer

2.1 or Oracle Web Application Server 3.0.

For more information, see the Release Notes provided with the ConText

Workbench or the "Oracle8 ConText Cartridge Workbench User's Guide".

----------------------------------------------------------------------------

TABLE OF CONTENTS

=================

INTRODUCTION

0.1 Purpose of this README

0.2 Cover Letter and Licensing

0.3 Oracle8 ConText Cartridge Documentation Currently Available

CHANGES TO CONTEXT OPTION

1.1 New Scoring Method for Text Queries

1.2 Processing of Stopwords in Text Queries Enhanced

1.3 Changes to ctxload Parameters

1.4 New Arguments for PL/SQL Stored Procedures and Functions

1.5 Supplied External Filters Installed/Upgraded with ConText

1.6 Enhanced Printjoin and Punctuation Handling for Indexing and Querying

1.7 Case-sensitive Thesaurus Expansion in Queries

1.8 Proper Names File (ctxprop.dmp) No Longer Provided

1.9 ctxsrvx Executable Obsolete

NEW FEATURES

2.1 Text Management (Loading, Storage, Filtering, etc.)

2.1.1 Case-sensitive Thesauri

2.1.2 Support for Startjoin and Endjoin Characters

2.1.3 Support for FTP in URL Data Store

2.1.4 Support for Document Update/Export in ctxload

2.1.5 Section and Section Group Definition

2.1.6 New Format for Master-Detail Tables

2.2 Indexing

2.2.1 Control Over Index Table Poplation During Indexing

2.2.2 Control Over DML Trigger Creation During Indexing

2.2.3 Case-sensitive Text Indexing

2.3 International

2.3.1 NLS Lexing

2.3.2 Enhanced Fuzzy Match for Non-English Languages and OCR Text

2.3.3 Chinese and Korean Lexers (Production Status)

2.3.4 Predefined Stoplists for Western European Languages

2.3.5 German Composite Text Indexing

2.4 Querying

2.4.1 Structured Data in In-memory Queries

2.4.2 Named Section Searching

2.4.3 Post-query Counting of Query Hits

2.4.4 Query Feedback Table

2.4.5 Case-sensitive Text Queries

2.4.6 NEAR Operator with Multiple Terms

2.5 Viewing

2.6 Linguistic

2.6.1 Sentence-level Gists and Theme Summaries

2.6.2 Improved Flexibility for Gist and Theme Summary Generation

2.6.3 Theme Highlighting

2.6.4 Improved Performance for Linguistics

2.6.5 Hierarchy Information for Theme Generation

2.7 ConText Workbench

2.7.1 Web-based Administration Tool

2.7.2 TextServer3 Migration Utility

2.7.3 Netscape Plug-in Viewer

NEW CONTEXT DATA DICTIONARY OBJECTS

3.1 New Tiles

3.2 New Predefined Preferences

3.3 New Template Policies

3.4 New Procedures and Functions

3.5 New Views

DOCUMENTATION CHANGES

4.1 Workbench User's Guide (NEW)

4.2 Administrator's Guide

4.3 Application Developer's Guide

INSTALLATION ISSUES

5.0 Location of ConText Installation Information

5.1 Perform Oracle8 and ConText Installation Separately

UPGRADE ISSUES

6.0 Requirements for Upgrading

6.1 Installer Upgrade Options for ConText 2.3.6

6.2 Manual Upgrade

6.3 Upgrading ConText Indexes

USAGE ISSUES

7.1 New Features Which Require Reindexing

7.2 Enabling One-Step Queries

7.2.1 Setting TEXT_ENABLE for All Users

7.2.2 Setting TEXT_ENABLE for the Session

7.3 Supplied External Filters

7.3.1 Installing the Filters

7.3.2 Using the Filters

7.4 Enabling the Linguistics

LINGUISTIC ISSUES

8.1 CTX_PROPER_NAMES Seed File No Longer Provided

8.2 Importing Knowledge Catalog/Lexicon Data

8.3 Starting ConText Servers for Theme Indexing and Querying

8.4 Specifying Settings for Linguistic Processing

8.5 Changing Settings for Theme Indexing and Queries

8.6 Obsolete Linguistic Setting Configurations

FIXES/ENHANCEMENTS

9.1 Fixes/Enhancements in Release 2.3

KNOWN LIMITATIONS

10.1 One-Step Queries in PL/SQL Not Supported (344159)

10.2 Indexing with Soundex Fails When Many Results Tables are

Present (401217)

10.3 One-Step Queries for LONG Columns in Remote Tables Not

Supported (436969)

10.4 Resume Index Creation/Optimization Not Supported (483325)

10.5 Scoring for NEAR Operator Inconsistent with Other Operators

-----------------------------------------------------------------------------

**********************

* *

* Introduction *

* *

**********************

0.1 Purpose of this README

--------------------------

This README file contains information about ConText, Release 2.3. It documents

any differences between ConText, Release 2.3, and its documented

functionality, as well as fixed bugs, and known problems and workarounds.

0.2 Cover Letter and Licensing

------------------------------

Please read the cover letter included with your Oracle8 package.

It contains important information about licenses for Oracle8 cartridges.

0.3 Oracle8 ConText Cartridge Documentation Available in this Release

-----------------------------------------------------------------

Oracle8 ConText Cartridge QuickStart

Oracle8 ConText Cartridge Administrator's Guide

Oracle8 ConText Cartridge Application Developer's Guide

Oracle8 Error Messages

Oracle8 ConText Cartridge Workbench User's Guide

-----------------------------------------------------------------------------

*********************************

* *

* Changes to ConText Option *

* *

*********************************

-------------------

NOTE: The following changes have been made from the previous release of

ConText. These changes may affect the functionality and operation of

the current release.

-------------------

1.1 New Scoring Method for Text Queries

=======================================

In previous releases, scoring was based on the number of occurrences of the

query term in the document. Each occurrence was assigned a default value

of 10, with the maximum score for a document set to 100.

In Release 2.3, a new scoring method is introduced. This method continues to

use number of occurrences, but includes document statistics, including the

number of documents in the text column and the total number of documents

returned in the query results, when calculating the overall score.

All of the operators, except for the NEAR (;) operator, which measures

distance between terms, function exactly as before.

In addition, the maximum score for a document is still 100; however, this

threshhold is not easily reached because the number of documents in which

a term occurs directly affects the number of occurrences required to

reach a score of 100.

-------------------

NOTE: DML affects the score results. In other words, a query for a specific

token on two different occasions may not return the same score for a

given document if DML has taken place between these 2 occasions.

In addition, the new scoring method does not affect scoring for

theme queries.

-------------------

1.2 Processing of Stopwords in Text Queries Enhanced

====================================================

In previous releases, if a stopword was specified as one of the operands

in a query expression, ConText returned an error and did not process the

query.

In Release 2.3, ConText replaces the stopword(s) with a NULL value and

processes the query to return results. No error is returned.

-------------------

NOTE: The effects of the replacement on the query varies depending on the

operators in the query expression.

-------------------

In most cases, the results returned by the query reflect the remaining operands

and operators in the expression. In a few select cases, such as query

expressions with a stopword as the first operand for the NOT (~) operator, no

results are returned.

The query feedback table, a new feature in Release 2.3, can be used to view

the replacement that ConText generates for queries that contain one or more

stopwords as operands.

1.3 Changes to ctxload Parameters

=================================

In Release 2.3, ctxload has been enhanced to provide the following

functionality:

-- document update and export, which enables users to update the

contents of the text column for a specified row or to

write the text column contents for the row to an

operating-system file.

-- import/creation of case-sensitive thesauri

To support these enhancements, the following command-line arguments have

been changed or added:

Release 2.0 argument and purpose Release 2.3 argument and purpose

-------------------------------- -------------------------------------

-file specified the name of a text -file same as before; however, it is

or thesaurus file to load, now also used to specify the

or a thesaurus export file name of the file from which

to create. text is loaded during document

update or the name of the file

created during document export.

-name specified the name of the -name same as before; however, it is

table to which text is loaded, now also used to specify the

the name of the thesaurus to policy name for the text

be created, or the name of the column to be updated or.

thesaurus to be exported. exported.

-export specifies that ctxload exports

the contents of the text

column for a row into an

operating-system file.

-update specifies that ctxload updates

the text column for a row with

the contents of a specified

operating-system file.

-pk specifies the primary key for

the row for which the text

column is updated/exported.

-thescase specifies whether the

thesaurus created by

ctxload is case-sensitive.

1.4 New Arguments for PL/SQL Stored Procedures and Functions

============================================================

The following new arguments have been added to the stored procedures and

functions in the ConText PL/SQL packages:

Package Procedure/Function New Argument Purpose

----------- ------------------- --------------- ---------------------------

CTX_DDL CREATE_INDEX create_trig Specify whether the DML

trigger is created during

indexing

pop_index Specify whether the ConText

index tables are populated

during indexing.

CTX_THES CREATE_THESAURUS case_sensitive Specify whether the thesurus

is case-sensitive.

CTX_QUERY OPEN_CON struct_query Include structured data in

an in-memory query.

CTX_LING REQUEST_GIST glevel Specify the type (paragraph or

sentenct) of Gist/theme

summary to create.

pov Specify whether to create a

Gist for a document or a theme

summary for one of the document

themes. If no value is

specified, a Gist and theme

summaries for all of the

document themes are created.

Each new argument has a default value. As a result, most of your applications

which call these procedures/functions will not require modification, provided

you want to used the defaults in the application.

-------------------

NOTE: Oracle Forms applications that call these procedures/functions will

require either valid or NULL values to be specified for each of the

new arguments.

-------------------

1.5 Supplied External Filters Installed/Upgraded with ConText

=============================================================

In previous releases, the external filters provided by Oracle were

distributed on a separate CD-ROM. These filters had to be manually

installed after installation of ConText. In addition, wrappers had to be

created for each external filter that you wanted to use.

In Release 2.3, the executables, as well as the wrappers, for the external

filters are installed automatically. If you upgrade from a previous

release, the old filters are upgraded to the new filters.

1.6 Enhanced Printjoin and Punctuation Handling for Indexing and Querying

=========================================================================

In previous releases, non-alphanumeric characters that appeared at the end of

a token and were defined as printjoins (attribute for the BASIC LEXER Tile)

were removed during indexing and queries. If the same printjoin character

appeared consecutively, all but one of the printjoins were removed.

In addition, the 'punctuations' attribute for the BASIC LEXER Tile

was not implemented. All non-alphanumeric characters that appeared at the end

of a token were simply removed during indexing and queries.

In Release 2.3, all printjoin characters, including consecutive printjoins, are

treated as alphanumeric characters and included in the index entry for the

token to which they are attached; however, if the printjoin character is also

defined as a punctuation character, the printjoin character is only treated as

an alphanumeric character if the character immediately following it is a

printjoin, skipjoin, or alphanumeric character.

For example, if the period (.) is defined as both a printjoin and a

punctuation character, the following transformations take place during

indexing and querying:

.doc --> .doc

dog.doc --> dog.doc

dog..doc --> dog..doc

dog. --> dog

dog... --> dog..

-------------------

NOTE: Because of the way the basic lexer identifies and stores tokens during

indexing, the change to printjoin handling has an effect on both

querying and highlighting of documents that have already been indexed.

You may want to drop and recreate your existing index(es) if you

defined printjoin characters for the BASIC LEXER Tile in your Lexer

preferences and any of the following conditions apply:

-- consecutive printjoins occur in the tokens in your documents

-- printjoins occur at the end of the tokens in your documents

-- same characters defined as **both** printjoins and

punctuations for the BASIC LEXER Tile

-------------------

1.7 Case-sensitive Thesaurus Expansion in Queries

=================================================

If a case-sensitive thesaurus is used in a query expansion, the thesaurus

lookup is performed using the query terms exactly as entered in the query.

For example, two queries for 'Turkey' and 'turkey', using a case-sensitive

thesaurus, may return two different expansions, whereas the same queries

using a case-insensitive thesaurus would return the same expansions.

-------------------

NOTE: To see the expansions performed using thesauri, use the query

feedback table, a new feature in Release 2.3, which provides detailed

information about the expansions that ConText performs in a query.

-------------------

1.8 Proper Names File (ctxprop.dmp) No Longer Provided

======================================================

The ctxprop.dmp file, which is used to populate the CTX_PROPER_NAMES table,

is no longer included in the ConText distribution.

The ConText Linguistics case-sensitivity routines no longer use the

CTX_PROPER_NAMES table to process all-uppercase and all-lowercase documents.

1.9 ctxsrvx Executable Obsolete

===============================

In previous releases, the Knowledge Catalog/Lexicon used by ConText to

generate theme indexes and linguistic output was compiled with the ctxsrv

executable. For users that did not require the linguistic capabilities of

ConText, a second executable, ctxsrvx, was provided. This executable was

compiled without the Knowledge Catalog/Lexicon and could be used to start

ConText servers with linguistics disabled.

With Release 2.3, the Knowledge Catalog/Lexicon is stored in a database table

owned by CTXSYS. Linguistic information is retrieved by ConText from the

table as needed.

As a result, the ctxsrvx executable is no longer needed. For backward

compatibility, ctxsrvx may still be provided; however, it is either a

link/shortcut to the ctxsrv executable or identical to ctxsrv.

-------------------

NOTE: As a result of the Knowledge Catalog/Lexicon now being stored in a

database table, to use the Linguistics, you must import the data

from the provided file into the database table after installation

or upgrade.

For more information, see "7.4 Enabling the Linguistics" in this

document.

-------------------

-----------------------------------------------------------------------------

********************

* *

* New Features *

* *

********************

ConText, Release 2.3, has a number of new features, designed to provide

more flexibility for data storage, indexing, and queries, as well as improve

query performance. The new features are grouped according to function:

2.1 Text Management (loading, storage, filtering, etc.)

2.2 Indexing

2.3 International

2.4 Querying

2.5 Viewing

2.6 Linguistics

2.7 ConText Workbench

For more information about the new text management, indexing, and

international features, see "Oracle8 ConText Cartridge Administrator's Guide".

For more information about the new querying, viewing, and Linguistic features,

see "Oracle8 ConText Cartridge Application Developer's Guide".

For more information about the new ConText Workbench features, see "Oracle8

ConText Cartridge Workbench User's Guide".

2.1 Text Management (Loading, Storage, Filtering, etc.) Features

================================================================

2.1.1 Case-sensitive Thesauri

-----------------------------

ConText supports creating case-sensitive thesauri through either

CTX_THES.CREATE_THESAURUS or the thesaurus load functionality in ctxload.

In a case-sensitive thesaurus, terms are stored exactly as entered. In a

case-insensitive thesaurus, terms are stored in all-uppercase, regardless of

the case in which they were entered.

-------------------

NOTE: Case-sensitive thesauri have an effect on query expansion. For

more information, see "1.7 Case-sensitive Thesaurus Expansion in

Queries".

-------------------

2.1.2 Support for Startjoin and Endjoin Characters

--------------------------------------------------

Through the new 'startjoin' and 'endjoin' attributes provided in the BASIC

LEXER Tile, users can specify characters that, when they appear at the

beginning and end of a token, explicitly start/end the token. In addition,

those characters specified as startjoins and endjoins are included in the

ConText index entry for the token.

Startjoins and endjoins are especially useful for defining document sections

to enable named section searching.

For more information about document sections and section searching, see

"2.1.5 Section and Section Group Definition" and "2.4.2 Named Section

"Searching

2.1.3 Support for FTP in URL Data Store

---------------------------------------

The URL data store (URL Tile) now supports the File Transfer Protocol (FTP)

in addition to HTTP and local file protocol.

URLs that use FTP, either on the Internet or on private intranets, can now be

accessed for indexing and viewing.

2.1.4 Support for Document Update/Export in ctxload

---------------------------------------------------

In previous releases, ctxload could only be used to create new rows and

insert new data into the text column for a given row. It could not be used

to update an existing row.

In Release 2.3, ctxload supports updating the text column for a row. In

addition, ctxload can be used to export the text column for a row to an

operating-system file.

2.1.5 Section and Section Group Definition

------------------------------------------

In Release 2.3, ConText supports restricting text queries to a user-defined

section. Sections are identified by user-defined start and end tags and are

grouped into section groups. A section group is attached to a text column

through the text indexing policy for the column and section information is

stored in the ConText index tables during indexing.

To enable named section searching, ConText provides new PL/SQL procedures

for creating and managing sections and section groups:

-- ADD_SECTION

-- CREATE_SECTION_GROUP

-- DROP_SECTION_GROUP

-- REMOVE_SECION

ConText also provides two new attributes for managing sections and section

groups:

-- 'section_group' (GENERIC WORD LIST Tile), which specifies the name of the

section group to be assigned to a text column

-- 'keep_tag' (HTML FILTER Tile), which specifies HTML tags to be retained by

the HTML filter for use in defining and identifying sections for HTML

documents.

Finally, Oracle provides a new template policy and new predefined preferences

for quickly and easily enabling basic section searching of HTML documents:

-- New template policy: TEMPLATE_BASIC_WEB

-- New predefined preferences:

-- BASIC_HTML_FILTER

-- BASIC_HTML_LEXER

-- BASIC_HTML_WORDLIST

For more information about section searching, see "2.4.2 Section Searching".

2.1.6 New Format for Master-Detail Tables

-----------------------------------------

In previous releases, ConText required you to create a policy on the detail

table in a master-detail relationship. This resulted in potentially slower

queries for text/themes and structured data, as well as slower DML and

non-intuitive one-step queries.

ConText, Release 2.3, supports defining a policy on the master table, rather

than the detail table. The detail table name and attributes are specified in

the Data Store preference, rather than in the policy.

To implement the new master/detail data store, a new Data Store Tile, MASTER

DETAIL NEW, is provided.

2.2 Indexing Features

======================

2.2.1 Control Over Index Table Poplation During Indexing

--------------------------------------------------------

ConText provides a new parameter, pop_index, for CTX_DDL.CREATE_INDEX. This

new parameter allows a user to create the ConText index tables for a policy

without generating the index. The ConText index tables can be populated

later through automatic DML, if enabled, or through manual DML

(CTX_DML.REINDEX).

2.2.2 Control Over DML Trigger Creation During Indexing

-------------------------------------------------------

ConText provides a new parameter, create_trig, for CTX_DDL.CREATE_INDEX. This

new parameter allows a user to create a ConText index, but supress creation

of the DML trigger that enables automatic DML for the index.

Updating of the index can then be controlled through manual DML

(CTX_DML.REINDEX).

2.2.3 Case-sensitive Text Indexing

----------------------------------

In previous releases, ConText automatically converted all tokens to

all-uppercase when creating text index entries. As a result, text queries

were always case-insensitive.

In Release 2.3, to support case-sensitive text queries, ConText provides a

new parameter, case_sensitive, for CTX_DDL.CREATE_INDEX. This new parameter

allows a user to create a ConText index in which the index entries are

recorded exactly as the tokens appear in the text.

2.3 International Features

==========================

2.3.1 NLS indexing

------------------

ConText supports the indexing of documents stored in any NLS

character set. Among other benefits, this feature enables the

indexing of documents based on the Unicode standard.

2.3.2 Enhanced Fuzzy Match for Non-English Languages and OCR Text

-----------------------------------------------------------------

Fuzzy matching is now available in German, French, Italian, Spanish,

and Dutch, as well as English.

In addition, ConText supports enhanced fuzzy matching for text that has been

scanned using Optical Character Recognition (OCR).

To enable fuzzy matching for these languages and OCR text, ConText provides

new values for the 'fuzzy_match' attribute (GENERIC WORD LIST).

2.3.3 Chinese and Korean Lexers (Production Status)

--------------------------------------------------

The Chinese and Korean lexers, which enable full text indexing in these

languages, have a status of full production in this release.

2.3.4 Predefined Stoplists for Western European Languages

----------------------------------------------------------

Stoplists, in the form of predefined Stoplist preferences, are provided for

the following Western European languages:

-- French

-- German

-- Italian

-- Spanish

2.3.5 German Composite Text Indexing

------------------------------------

To support querying for root nouns in German-language text and retrieving

text that contains either the root nouns or any compound nouns in which the

roots appear, the text lexer used for indexing German-language text (BASIC

LEXER) now provides an attribute named 'composite' to enable decompounding of

compound nouns. If enabled, this feature separates compound nouns into their

root nouns and stores the roots, as well as the original compound nouns, in

the text index.

-------------------

NOTE: German composite indexing does not support case-sensitive indexing,

another new feature in 2.3.6. Tokens in a German composite index

are always stored in all-uppercase.

-------------------

2.4 Querying Features

=====================

2.4.1 Structured Data in In-memory Queries

------------------------------------------

In addition to text search criteria, application developers can include

structured data search criteria in in-memory queries.

The CTX_QUERY.OPEN_CON procedure provides a new parameter, struct_query, to

enable structured data criteria in in-memory queries.

2.4.2 Named Section Searching

-----------------------------

ConText now supports restricting a text query to a specified, user-defined

section.

One or more sections are created by the text administrator and assigned to a

section group, which is then assigned to a policy for a text column. During

creation of the ConText index, the section information for each token is

stored in the index along with the token.

Using the new WITHIN operator, text queries against the column can be

restricted to one of the sections for the column.

For more information about defining sections and section groups, see

"2.1.5 Section and Section Group Definition".

2.4.3 Post-query Counting of Query Hits

---------------------------------------

In previous releases, the only method of counting the number of hits in a

two-step or in-memory query was to first call CTX_QUERY.COUNT_HITS to return

the number of hits, then submit the query to return the results.

In Release 2.3, a new function, CTX_QUERY.COUNT_LAST, has been added, which

returns a count of the number of rows from the last two-step or in-memory

query.

2.4.4 Query Feedback Table

--------------------------

In Release 2.3, ConText provides a new procedure, CTX_QUERY.FEEDBACK, which

generates information about the expansions, if any, that occur in a query, as

well as the precedence of the operators and operands in the query, and stores

the information in a database table. This query feedback table can then be

queried using standard SQL to return the information generated about the query.

In general, CTX_QUERY.FEEDBACK is run prior to submitting a ConText query

so that the information can be used to refine the query before the query is

submitted.

For example, the query feedback can be used to inform the user entering the

query that their current query will be expanded to include a large number of

terms and whether the user wants the query to be submitted as entered.

2.4.5 Case-sensitive Text Queries

---------------------------------

In Release 2.3, if a case-sensitive text index is created for a column, text

queries on the index are case-sensitive, resulting in higher precision.

For example, in previous releases, a text query for "Oracle" returned

all documents that contained any occurrences of the term "oracle", "Oracle",

"ORACLE", etc.

In this release, a text query for "Oracle" on a case-sensitive index returns

only the documents that contain the term "Oracle".

2.4.6 NEAR Operator with Multiple Terms

---------------------------------------

In Release 2.3, the NEAR operator can be used to score retrieved documents

based on the proximity of two or more terms.

2.5 Viewing Features

====================

See "2.6.3 Theme Highlighting" and "2.7.3 Netscape Plug-in Viewer" for more

information about the new document viewing features in Release 2.3.

2.6 Linguistic Features (English-text Only)

===========================================

-------------------

NOTE: Beginning with this release, "generic Gists" are referred to as

"Gists" and "point-of-view (POV) Gists" are referred to as

"theme summaries".

-------------------

2.6.1 Sentence-level Gists and Theme Summaries

----------------------------------------------

In Release 2.3, application developers can choose to generate either

paragraph-level (DEFAULT) or sentence-level Gists/theme summaries.

Paragraph summarization selects the most representative paragraphs in a

document to build Gists and theme summaries. Sentence-level summarization

selects the most representative sentences.

The CTX_LING.REQUEST_GIST procedure provides a new parameter, glevel, to

enable paragraph-level and sentence-level Gists.

2.6.2 Improved Flexibility for Gist and Theme Summary Generation

----------------------------------------------------------------

In Release 2.3, application developers have the following options when

requesting Gists and theme summaries:

1) document Gist and theme summaries for each document theme (DEFAULT)

2) document Gist only

3) theme summary for specified theme only

The CTX_LING.REQUEST_GIST procedure provides a new parameter, pov, which can

be used to limit Gist/theme summary generation to the document Gist only or

to a theme summary for a specific theme in the document.

2.6.3 Theme Highlighting

------------------------

In Release 2.3, the CTX_QUERY.HIGHLIGHT procedure now supports generating

highlight information for a specified theme in a specified document. With

theme highlighting, the paragraphs that contribute to the theme for a

document are identified and markup information or the actual marked-up text

is stored in output tables specified by the user.

2.6.4 Improved Performance for Linguistics

------------------------------------------

In Release 2.3, the performance of the Linguistics has been greatly improved.

In addition, the ctxsrv executable has been reduced in size by approximately

80%, resulting in less memory required for Linguisitic processing.

2.6.5 Hierarchy Information for Theme Generation

------------------------------------------------

In previous releases, CTX_LING.REQUEST_THEMES returned only the themes

for a document and stored each theme as a row in a user-specified table.

In Release 2.3, if a document theme has an entry in the ConText Knowledge

Catalog, REQUEST_THEMES returns the parent concept for the theme, as well as

all the other higher-level concepts for the theme, as stored in the Knowledge

Catalog. In effect, REQUEST_THEMES returns the full Knowledge Catalog

hierarchy of concepts for each document theme.

A theme and its hierarchy is stored as a single string, separated by

commas (:). The top level concept in the hierarchy occurs first in the

string, followed by each successively lower-level concept, with the actual

theme occurring last.

In addition, ConText provides two new procedures, CTX_LING.SET_FULL_THEMES

and CTX_LING.GET_FULL_THEMES, for managing the theme hierarchies returned by

REQUEST_THEMES.

2.7 ConText Workbench Features

==============================

2.7.1 Web-based Administration Tool

-----------------------------------

In previous releases, the GUI System Administration tool was only available

for the Windows NT and Windows 95 platforms

An HTML version of the System Administration tool is now provided for managing

ConText from any Web-enabled environment.

2.7.2 TextServer3 Migration Utility

-----------------------------------

A new utility is provided for migrating a TextServer3 data dictionary to a

ConText data dictionary.

2.7.3 Netscape Plug-in Viewer

-----------------------------

In Release 2.3, the ConText Workbench provides a plug-in for viewing documents

with query terms highlighted within a Netscape browser.

For Microsoft Word (v6.0 and earlier) and WordPerfect (v6.0 and earlier)

documents, the Netscape plug-in provides WYSIWYG viewing of the documents.

For those document formats for which ConText supports filtering, but not

WYSIWYG viewing, the viewer converts the documents to ASCII before displaying.

-----------------------------------------------------------------------------

*******************************************

* *

* New ConText Data Dictionary Objects *

* *

*******************************************

-------------------

NOTE: For a list of the new arguments provided for exisiting procedures and

functions, see "1.4 New Arguments for PL/SQL Stored Procedures and

Functions".

-------------------

3.1 New Tiles

=============

ConText, Release 2.3, provides a new Data Store Tile, MASTER DETAIL NEW, to

support the new method for defining master-detail relationships between

tables.

3.2 New Predefined Preferences

==============================

ConText, Release 2.3, provides the following new preferences for quickly and

easily enabling basic section searching in HTML documents:

-- BASIC_HTML_FILTER

-- BASIC_HTML_LEXER

-- BASIC_HTML_WORDLIST

3.3 New Template Policies

=========================

ConText, Release 2.3, provides a new template policy, TEMPLATE_BASIC_WEB, that

can be used to create section-enabled policies for columns that contain URLs to

HTML documents.

It uses the URL Data Store preference and the new predefined preferences (see

above).

3.4 New Procedures and Functions

================================

ConText, Release 2.3, provides new PL/SQL procedures and functions to enable

the following new features:

-- Named Section Searching:

-- CTX_DDL.ADD_SECTION

-- CTX_DDL.CREATE_SECTION_GROUP

-- CTX_DDL.DROP_SECTION_GROUP

-- CTX_DDL.REMOVE_SECION

-- Query Feedback: CTX_QUERY.FEEDBACK

-- Query Hits Counting: CTX_QUERY.COUNT_LAST

-- Hierarchy Information for Theme Generation:

-- CTX_LING.SET_FULL_THEMES

-- CTX_LING.GET_FULL_THEMES

3.5 New Views

=============

ConText, Release 2.3, provides the following new views for monitoring

sections and section groups:

-- CTX_ALL_SECTION GROUPS

-- CTX_ALL_SECTIONS

-- CTX_USER_SECTION_GROUPS

-- CTX_USER_SECTIONS

CTX_ALL_SECTION_GROUPS displays the ID, name, and owner of all the

section groups that have been created.

CTX_ALL_SECTIONS displays the ID, name, owner, and section definition

(start tags, end tags, etc.), for each section that has been created.

It also displays the ID for the section group to which the section belongs.

CTX_USER_SECTION_GROUPS displays the name of all the section groups that

have been created for the current user.

CTX_USER_SECTIONS displays the ID, name, and section definition (start

tags, end tags, etc.), for each section that has been created for the

current user. It also displays the ID for the section group to which the

section belongs.

-----------------------------------------------------------------------------

*****************************

* *

* Documentation Changes *

* *

*****************************

4.1 Workbench User's Guide (**NEW**)

====================================

In Release 2.3, a new manual is included in the ConText documentation set.

The ConText Workbench user's Guide serves as a single, complete source of

setup and usage information for all the components that comprise the Oracle

ConText Option Workbench.

It combines the descriptions and usage information for existing Workbench

components (collected from various other ConText documentation) with the

descriptions, setup information, and usage information for the new components

in the ConText Workbench, Release 2.3:

-- Administration tools

-- System Administration for Windows NT and 95

-- Configuration Manager for Web environments (**NEW**)

-- Sample Oracle Forms application

-- Document viewer

-- custom control

-- Netscape plug-in (**NEW**)

-- Input/Output Utility

4.2 Administrator's Guide

=========================

-- ConText, Release 2.3, new features documented

-- Chapters removed:

-- "Chapter 7, Configuring ConText Option"

-- "Chapter 8, Tuning ConText Option"

-------------------

NOTE: These chapters will be available in a new manual, "Oracle

ConText Option Tuning", which will be released after

Release 2.3.

-------------------

-- Reference section ("Part IV") removed and chapters reorganized into

relevent section:

-- "Chapter 9, Executables and Utilities" split into two chapters:

-- "ConText Server Executables and Utilities" in "Part I"

-- "Text Loading Utility" in "Part II"

-- "Chapter 10, Data Dictionary" merged into "Understanding the ConText

Data Dictionary" chapter

-- "Chapter 11, PL/SQL Packages" split into two chapters:

-- "PL/SQL Packages - Administration" in "Part I"

-- "PL/SQL Packages - Text Management" in "Part II"

-- "Appendix A, ConText Server and Queue Views" and "Appendix B, ConText Data

Dictionary Views" combined into a single appendix named "ConText Views"

-- New appendix named "Supplied Stoplists", which contains a list of the stop

words for each of the predefined Stoplist preferences.

4.3 Application Developer's Guide

=================================

-- ConText, Release 2.3, new features documented

-- New Chapters:

-- "Chapter 5: Query Expression Feedback"

-- "Appendix C: Stopword Transformations"

-- Chapters removed:

-- "Chapter 1: Introduction" removed from manual

-- "Appendix C: Forms Sample Application" moved to new manual, "Oracle

ConText Option Workbench User's Guide"

-- "Appendix D: Input/Output Utility" moved to new manual, "Oracle8

ConText Cartridge Workbench User's Guide"

-- Chapters enhanced:

-- "Chapter 3, Query Methods" (renumbered to "Chapter 2") expanded

to include sections on counting query hits.

-- "Chapter 5, Theme Queries" (renumbered to "Chapter 4") expanded

and reorganized into three sections:

-- "Understanding Theme Queries"

-- "Constructing Theme Queries"

-- "Refining Theme Queries"

-- "Chapter 6, Viewing Highlighted Text" renamed to "Chapter 6:

Document Presentation" to reflect the addition of new document

presentation features for Release 2.3

-----------------------------------------------------------------------------

***************************

* *

* Installation Issues *

* *

***************************

5.0 Location of ConText Installation Information

================================================

The information required for installing ConText is located in the Oracle8

installation documentation for your operating system.

Additional installation information for ConText may be located in the

Oracle8 Server release notes for your operating system.

5.1 Perform Oracle8 and ConText Installation Separately

=======================================================

Because the ConText data dictionary should be installed into an existing

tablespace, Oracle Corporation suggests installing ConText in a separate step

after Oracle8 has been installed and the appropriate tablespaces have been

created.

After Oracle8 is installed and the appropriate tablespaces are created, return

to the Installer screen which lists the available produts for installation and

select the ConText entry.

-----------------------------------------------------------------------------

**********************

* *

* Upgrade Issues *

* *

**********************

6.0 Requirements for Upgrading

==============================

Upgrading from a previous version of ConText must occur in two separate steps:

1) Upgrade/migrate your Oracle database to the version 8.0.4

2) Upgrade ConText from the previous version to version 2.3.6

-------------------

NOTE: Version 8.0.4 of Oracle is the only supported Oracle database version

for version 2.3.6 of ConText.

-------------------

In addition, all ConText servers that are currently running must be shut down

BEFORE upgrade.

For more information about upgrading ConText, see the Oracle8 Server

installation documentation for your operating system.

6.1 Installer Upgrade Options for ConText 2.3.6

===============================================

When upgrading to ConText, Release 2.3, the Oracle Installer provides the

following choices.

Migration:

-- 1.1.2 (7.3.2) to 2.3.6 (8.0.4)

-- 2.0.4 (7.3.3) to 2.3.6 (8.0.4)

-- 2.3.4 (7.3.4) to 2.3.6 (8.0.4)

Upgrade:

-- 2.0.6 (8.0.3) to 2.3.6 (8.0.4)

6.2 Manual Upgrade

==================

If you are upgrading manually, you must copy the new software (executables,

scripts, packages, etc.) to the appropriate subdirectories in the ConText

directory, then run the appropriate scripts for your version of ConText:

1.1.2 to 2.0.4 --> sys20000.sql as SYS

upg20000.sql as CTXSYS

2.0.4 to 2.3.4 --> upg23400.sql as CTXSYS

2.3.4 to 2.3.6 --> **NONE** (no database changes)

-------------------

NOTE: After manual upgrade from 1.1.2, if you want to use the ConText

Linguistics, you must import the ctxset.dmp file, using the Import/

Export utility provided by Oracle:

IMP ctxsys/<passwd> file=ctxset.dmp full=y ignore=y

The ctxset.dmp file is generally located in the admin subdirectory in

the ConText directory in your Oracle home directory; however, the

exact location is operating system dependent.

For more information about the exact location of the file, see the

Oracle8 Server documentation specific to your operating system.

-------------------

6.2 Upgrading ConText Indexes

=============================

After upgrading to Release 2.3 from a previous release, you **MUST** call the

CTX_DDL.UPGRADE_INDEX procedure once for each existing ConText index.

UPGRADE_INDEX converts ConText indexes to the current release.

UPGRADE_INDEX can be called by the owner of the policy for the index or by

CTXSYS.

In the following example, the ConText index for the 'html_docs' policy is

upgraded to version 2.3 by the policy owner:

exec ctx_ddl.upgrade_index('html_docs')

For more information about CTX_DDL.UPGRADE_INDEX, see "Oracle8 ConText

Cartridge Administrator's Guide".

-----------------------------------------------------------------------------

********************

* *

* Usage Issues *

* *

********************

7.1 New Features Which Require Reindexing

=========================================

To use the following new features or enhancements, you must drop existing

ConText indexes on the text columns for which you want to use the new feature,

then create new ConText indexes for the columns:

-- named section searching

-- startjoin and endjoin characters

-- new master/detail data store method

-- enhancements for printjoin and punctuation characters

-- case-sensitive text indexing

-- German composite text indexing

-------------------

NOTE: In addition to creating new ConText indexes, you also may have to

perform some setup to use these features.

For example, to use section searching for a text column, you must

create a section group, create sections for the section group, and

assign the section group to the policy for the text column. Then you

can create the ConText index for the column.

-------------------

7.2 Enabling One-Step Queries

=============================

If you want to use one-step queries in ConText, you must set the ConText

initialization parameter TEXT_ENABLE to TRUE. TEXT_ENABLE enables Oracle8

to recognize the CONTAINS SQL function utilized in one-step queries.

You can set TEXT_ENABLE for all users and sessions in the initsid.ora file.

You can also set TEXT_ENABLE for the current session using the SQL command,

ALTER SESSION.

-------------------

NOTE: Once you have set TEXT_ENABLE, start one or more ConText servers

with the Query (Q) personality to ensure one-step queries are

processed.

-------------------

7.2.1 Setting TEXT_ENABLE for All Users

---------------------------------------

To set TEXT_ENABLE for all users, perform the following steps:

1. Shut down the database.

2. Add the following line to the initsid.ora file:

text_enable=true

-------------------

NOTE: Setting TEXT_ENABLE for all users does not create a security

violation. It only ensures that one-step text queries can be

processed.

-------------------

7.2.2 Setting TEXT_ENABLE for the Session

-----------------------------------------

To set TEXT_ENABLE for the current session only, issue the following SQL

command:

alter session set text_enable = true

7.3 Supplied External Filters

=============================

ConText, Release 2.3, supports using external filters to filter text for

creating ConText indexes and highlighting. External filters are provided by

ConText for a number of formats, including:

-- PDF

-- AmiPro

-- Lotus 1-2-3

-- Microsoft Excel

-- Microsoft PowerPoint

-- Microsoft Word

7.3.1 Installing the Filters

----------------------------

The external filters are supplied as executables which are automatically

installed with ConText. In addition, wrappers for each of the executables are

installed with ConText. The location of the filter executables and wrappers

is dependent on your operating system.

For more information about the location of the filter executables and

wrappers, see the Oracle8 installation documentation specific to your

operating system.

-------------------

NOTE: In previous releases, the supplied external filters were included on

a separate CD-ROM and had to be installed manually.

In this release, the installation of the supplied external filters is

handled by the Oracle Installer.

-------------------

7.3.2 Using the Filters

-----------------------

To use one or more external filters for a text column, create a Filter

preference that calls the wrapper for each desired filter, then use the

preference in the policy for the column.

-------------------

NOTE: The type of column (single-format or mixed-format) determines the Tile

and attributes that you must set when creating the Filter preference.

If the column stores documents in a single format, use the USER FILTER

Tile and the 'command' attribute.

If the column stores documents in more than one format, use the

BLASTER FILTER Tile and the 'executable' attribute.

-------------------

For more information about creating Filter preferences for external filters,

see "Oracle8 ConText Cartridge Administrator's Guide".

7.3.3 PDF Filter Limitations

----------------------------

The PDF filter provided for all platforms other than Windows NT and Sun Solaris

2.x has a status of BETA. The BETA PDF filter does not support multi-column

documents or documents over 1 Mb in size.

For Windows NT and Sun Solaris 2.x, the PDF filter has a status of production.

The filter correctly processes multi-column and large PDF documents; however,

the filter replaces bullet marks with characters.

7.4 Enabling the Linguistics

============================

In Release 2.3, the Knowledge Catalog/Lexicon utilized by the ConText

Linguistics is stored in a database table owned by CTXSYS, rather than

compiled as part of the ctxsrv executable.

However, the table is not populated during installation/upgrade of ConText.

If you want to create theme indexes, perform theme queries, or generate

linguistic output for your English text, the data for the Knowledge Catalog

table must be imported after installation/upgrade from the ctxlex.dmp export

file provided by ConText.

The following procedure provides instructions for importing the data:

1) Ensure the tablespace for CTXSYS has at least 30-40 Megabytes available.

2) Create a rollback segment in a tablespace that has at least 30 Megabytes

available. For example:

sql> create rollback segment rollbk_temp tablespace drsys storage (

2 initial 5M next 5M optimal 10M minextents 2 maxextents unlimited);

This example creates a rollback segment named 'rollbk_temp' in a tablespace

named 'drsys'.

3) Bring the new rollback segment online and bring all existing rollback

segments, **EXCEPT** system offline.

4) Use the IMP utility to import the ctxlex.dmp file as CTXSYS:

IMP ctxsys/<passwd> file=ctxlex.dmp full=y ignore=y buffer=50000 feedback=50000

5) Bring the new rollback segment offline and bring all existing rollback

segments online.

-------------------

NOTE: The ctxlex.dmp file is generally located in the admin subdirectory in

the ConText directory in your Oracle home directory.

For example, in a UNIX-based environment, the file is located in

$ORACLE_HOME/ctx/admin. In a Windows NT environment, the file is

located in ORACLE_HOME\CTX80\ADMIN.

For more information about the exact location of the file, see the

Oracle8 Server documentation specific to your operating system.

-------------------

-----------------------------------------------------------------------------

*************************

* *

* Linguistic Issues *

* *

*************************

8.1 CTX_PROPER_NAMES Table Obsolete

===================================

In Release 2.3, the CTX_PROPER_NAMES is no longer used by the ConText

Linguistics to process English-text documents that are in all-uppercase

or all-lowercase.

As a result, the ctxprop.dmp file is no longer provided. The ctxprop.dmp

file contained a list of proper names that, when imported into the

CTX_PROPER_NAMES table, enabled the ConText Linguistics case-sensitivity

routines.

If you are upgrading from a previous release and have already imported

ctxprop.dmp into this table, you can delete or truncate the table.

8.2 Importing Knowledge Catalog/Lexicon Data

============================================

Before ConText can be used to create theme indexes or generate linguistic

output, the Knowledge Catalog/Lexicon data must be imported into the lexicon

table.

For more information, see "7.4 Enabling the Linguistics".

8.3 Starting ConText Servers for Theme Indexing and Querying

============================================================

Unlike Gist, theme, and theme summary generation, theme indexing and theme

querying do not require ConText servers running with the Linguistics (L)

personality.

Creation of theme indexes is a DDL operation performed by ConText servers

running with the DDL (D) personality. Similarly, theme queries are processed

by ConText servers running with the Query (Q) personality.

For more information about ConText servers, personalities, and the

ctxsrv executable, see "Oracle8 ConText Cartridge Administrator's

Guide".

8.4 Specifying Settings for Linguistic Processing

=================================================

Both the Linguistics and the Theme Lexer use linguistic settings, specified

as setting configurations, to determine:

1) processing for all-uppercase and all-lowercase text or mixed-case text

2) size of theme summaries

3) size and generation method for Gists

ConText provides two predefined setting configurations:

-- GENERIC (mixed-case text)

-- SA (all-uppercase or all-lowercase text)

'GENERIC' is the default predefined setting configuration and is automatically

enabled for each ConText server at start up.

You can create your own custom setting configurations using either of the GUI

administration tools provided in the ConText Workbench. You create custom

setting configurations using the predefined setting configurations as

templates.

You can enable the 'SA' setting configuration or a custom setting

configuration for the current database session/user, using the

SET_SETTINGS_LABEL in the CTX_LING PL/SQL package.

-------------------

NOTE: Successful theme and Gist generation relies on specifying a setting

configuration that matches the size and type of documents to be

processed.

However, theme indexing and theme queries do not rely as heavily on

the settings. As such, the default setting configuration should be

used always for theme indexing/queries.

-------------------

For more information about setting configurations and using the linguistic

settings, see "Oracle8 ConText Cartridge Application Developer's Guide".

8.5 Changing Settings for Theme Indexing and Queries

====================================================

Because theme indexes are generated on a per column basis and must be

regularly updated/maintained to ensure accurate results for theme queries and

SET_SETTINGS_LABEL only enables a setting configuration for the current

session/user, Oracle Corporation does **NOT** recommend changing the

linguistic settings for theme indexing/queries.

-------------------

NOTE: Once a theme index has been created for a column, the setting

configuration for the current session and any subsequent sessions

should **NOT** be changed.

-------------------

For theme and Gist/theme summary generation, which is performed on a per

document basis and generally requires manual maintenence, enabling a different

setting is not necessary, but is recommended if the default setting

configuration does not generate the desired linguistic output.

8.6 Obsolete Predefined Setting Configurations

==============================================

The following predefined setting configurations are obsolete and provided only

for backward compatibility:

-- S

-- P

-- PP

-- PS

-- PSP

-- PSA

-- PSAP

If you use any of these obsolete setting configurations in an existing

application, you can continue to use them without any adverse effects; however,

for new applications, you should only use the 'GENERIC' or 'SA' or any custom

setting configurations that you have defined.

-----------------------------------------------------------------------------

**************************

* *

* Fixes/Enhancements *

* *

**************************

9.1 Fixes/Enhancements in Release 2.3

=====================================

403354

Oracle Installer hangs when creating database objects for CTXSYS.

405408

Error during ConText index creation with Soundex enabled due to incorrect

format for 'stclause' and 'instclause' attributes in GENERIC WORD LIST Tile.

451530

Queries only return hits from master-detail tables if search terms are

encountered in the first 2000 characters of the text.

458969

Two-step query with large query ID gives incorrect value in results table.

459036

ctxsrvx core dumps when NLS_LANG is set to AL24UTFFSS

466374

A cursor is left open when executing a one-step query in which the query

expression contains a user-defined function.

474426

Performance problems with struct_query in a two-step query if the datatype for

the textkey is not CHAR or VARCHAR2.

474471

OPTIMIZE_INDEX uses the STORAGE parameters for a view, rather than the STORAGE

parameters specified in the Engine preference.

476799

Fuzzy search with master-detail tables crashes ConText servers.

478063 (Also fixes 462571)

Access control checks user rather than effective user.

481502

Internal error encountered, user unable to create a ConText index on a text

table as the owner.

495711

Incorrect results when ID in CTX_QUERY.HIGHLIGHT is 10 digits or greater.

500035

After loading large amounts of data and calling SYNC, nothing appears in the

DRQ_INPROG table and ConText hangs.

504488

Poor performance for two-step queries that include structured data

505982

Cannot drop index when server started and there is pending DML.

509331

Query length over 2000 characters doesn't return error 'Query too complex'

when expected.

510916

Encounter Oracle error --- ORA-04098: TRIGGER 'T1_TR_ROW' IS INVALID AND

FAILED RE-VALIDATION --- when trying to insert into a text table when a

ConText index has been dropped on another text table.

512209

Internal errors encountered when generating themes and gists with user

defined settings.

512698

Don't remove duplicates from DRQ_PENDING for better performance on large

DML batches.

512699

Fold DRQ_INPROG into DRQ_PENDING for better performance on batch DML.

514404

Cannot create index on theme policy when record contains policy or preference.

517274

Cannot load documents larger in size than 64K when using automated text loading

(sources and ctxload).

520688 (Also fixes 487974)

HTML filter not indexing some keywords.

521663

Internal error while executing fuzzy match on term, when fuzzy match returns

large number of expansions.

532547

Internal error encountered in ConText log when INSERT and DELETE text using

multiple servers.

569521

ConText query performance degrades as the number of database users

increases.

-----------------------------------------------------------------------------

*************************

* *

* Known Limitations *

* *

*************************

The following section lists known problems that may affect your usage of

ConText, Release 2.3.

10.1 One-Step Queries in PL/SQL Not Supported (344159)

======================================================

ConText, Release 2.3, does not support using one-step queries within

PL/SQL blocks.

If you are using PL/SQL to develop applications, use two-step queries

(CTX_QUERY.CONTAINS) or dynamic SQL (DBMS_SQL package).

The following example illustrates using dynamic SQL in a PL/SQL block to fetch

the score returned by a one-step query for documents containing the word

'test'.

declare

myscore number;

c1 number;

n number;

f number;

stmt varchar2(1024);

begin

stmt := 'select score(0) from docs where contains(text,'test')>0';

c1 := dbms_sql.open_cursor;

dbms_sql.parse(c1, stmt, dbms_sql.v7);

dbms_sql.define_column(c1, 1, myscore);

n := dbms_sql.execute(c1);

loop

f := dbms_sql.fetch_rows(c1);

exit when f = 0;

dbms_sql.column_value(c1, 1, myscore);

dbms_output.put_line('score : ' || myscore);

end loop;

dbms_sql.close_cursor(c1);

end;

10.2 Indexing with Soundex Fails When Many Results Tables are Present (401217)

==============================================================================

If a large number of internal result tables (utilized in one-step queries)

are present during text indexing with Soundex enabled, the index creation will

fail.

-------------------

Suggestion: Before performing text indexing with Soundex, verify that all

unnecessary result tables have been dropped.

-------------------

10.3 One-Step Queries for LONG Columns in Remote Tables Not Supported (436969)

==============================================================================

ConText, Release 2.3, does not support one-step queries on LONG and LONG RAW

columns in remote database tables and views.

10.4 Resume Index Optimization Not Supported

============================================

If index creation or optimization fails during processing,

CTX_DDL.RESUME_FAILED_INDEX can be used to resume the operation from the

failure point; however, additional errors may occur for index optimization.

-------------------

Suggestion: If index optimization fails during processing, do not use

RESUME_FAILED_INDEX. Instead, fix the reason for the error,

clear the errored request from the error logs, and start

optimization from the beginning.

-------------------

10.5 Scoring for NEAR Operator Inconsistent with Other Operators

================================================================

All the operators in a text query produce scores for query hits based on the

frequency of the occurrences of the query terms, with the exception of the

NEAR operator.

When the NEAR operator is used in a text query, it produces a

score based on the proximity of the two query terms. If the two terms are

adjacent, the score is 100. As the number of tokens between the two terms

increases, the score decreases.

As such, the two scoring methods are inconsistent and, when used together,

will not generally produce meaningful scores.

To avoid this situation, Oracle Corporation suggests that you do not combine

usage of the NEAR operator with other operators.

-----------------------------------------------------------------------------