10
CTX_QUERY Package

This chapter describes the CTX_QUERY PL/SQL package you can use for generating query feedback, counting hits, and creating stored query expressions.

The CTX_QUERY package includes the following procedures and functions:

Name	Description
BROWSE_WORDS	Returns the words around a seed word in the index.
COUNT_HITS	Returns the number hits to a query.
EXPLAIN	Generates query expression parse and expansion information.
HFEEDBACK	Generates hierarchical query feedback information (broader term, narrower term, and related term).
REMOVE_SQE	Removes a specified stored query expression from the SQL tables.
STORE_SQE	Executes a query and stores the results in stored query expression tables.

---

BROWSE_WORDS

This procedure enables you to browse words in an Oracle Text index. You specify a seed word and BROWSE_WORDS returns the words around it in the index, and a rough count of the number of documents that contain each word.

This feature is useful for refining queries. You can identify the following:

• unselective words (words that have low document count)
• misspelled words in the document set

Syntax 1: To Store Results in Table

ctx_query.browse_words( 


index_name  IN   VARCHAR2, 
seed        IN   VARCHAR2, 
restab      IN   VARCHAR2, 
browse_id   IN   NUMBER   DEFAULT 0, 
numwords    IN   NUMBER   DEFAULT 10, 
direction   IN   VARCHAR2 DEFAULT BROWSE_AROUND,
part_name   IN   VARCHAR2 DEFAULT NULL

);

Syntax 2: To Store Results in Memory

ctx_query.browse_words( 


index_name  IN      VARCHAR2, 
seed        IN      VARCHAR2, 
resarr      IN OUT  BROWSE_TAB, 
numwords    IN      NUMBER   DEFAULT 10, 
direction   IN      VARCHAR2 DEFAULT BROWSE_AROUND,
part_name   IN      VARCHAR2 DEFAULT NULL 

);

index

Specify the name of the index. You can specify schema.name. Must be a local index.

seed

Specify the seed word. This word is lexed before browse expansion. The word need not exist in the token table. seed must be a single word. Using multiple words as the seed will result in an error.

restab

Specify the name of the result table. You can enter restab as schema.name. The table must exist before you call this procedure, and you must have INSERT permissions on the table. This table must have the following schema.

Column	Datatype
browse_id	number
word	varchar2(64)
doc_count	number

Existing rows in restab are not deleted before BROWSE_WORDS is called.

resarr

Specify the name of the result array. resarr is of type ctx_query.browse_tab.

type browse_rec is record (
   word varchar2(64),
   doc_count number
);
type browse_tab is table of browse_rec index by binary_integer;

browse_id

Specify a numeric identifier between 0 and 2³². The rows produced for this browse have a value of in the browse_id column in restab. When you do not specify browse_id, it defaults to 0.

numwords

Specify the number of words returned.

direction

Specify the direction for the browse. You can specify one of:

value	behavior
BEFORE	Browse seed word and words alphabetically before the seed.
AROUND	Browse seed word and words alphabetically before and after the seed.
AFTER	Browse seed word and words alphabetically after the seed.

Symbols CTX_QUERY.BROWSE_BEFORE, CTX_QUERY.BROWSE_AROUND, and CTX_QUERY.BROWSE_AFTER are defined for these literal values as well.

part_name

Specify the name of the index partition to browse.

Example

Browsing Words with Result Table

begin
ctx_query.browse_words('myindex','dog','myres',numwords=>5,direction=>'AROUND');
end;

select word, doc_count from myres order by word;

WORD       DOC_COUNT
--------   ----------
CZAR       15
DARLING    5
DOC        73
DUNK       100
EAR        3

Browsing Words with Result Array

set serveroutput on;
declare
  resarr ctx_query.browse_tab;
begin
ctx_query.browse_words('myindex','dog',resarr,5,CTX_QUERY.BROWSE_AROUND);
for i in 1..resarr.count loop
  dbms_output.put_line(resarr(i).word || ':' || resarr(i).doc_count);
end loop;
end;

---

COUNT_HITS

Returns the number of hits for the specified query. You can call COUNT_HITS in exact or estimate mode. Exact mode returns the exact number of hits for the query. Estimate mode returns an upper-bound estimate but runs faster than exact mode.

Syntax

CTX_QUERY.COUNT_HITS ( 
    index_name  IN VARCHAR2, 
    text_query  IN VARCHAR2, 
    exact       IN BOOLEAN  DEFAULT TRUE,
    part_name   IN VARCHAR2 DEFAULT NULL
) RETURN NUMBER; 

index_name

Specify the index name.

text_query

Specify the query.

exact

Specify TRUE for an exact count. Specify FALSE for an upper-bound estimate. Specifying FALSE returns a less accurate number but runs faster.

part_name

Specify the name of the index partition to query.

Notes

If the query contains structured criteria, you should use SELECT COUNT(*).

---

EXPLAIN

Use CTX_QUERY.EXPLAIN to generate explain plan information for a query expression. The EXPLAIN plan provides a graphical representation of the parse tree for a Text query expression. This information is stored in a result table.

This procedure does not execute the query. Instead, this procedure can tell you how a query is expanded and parsed before you issue the query. This is especially useful for stem, wildcard, thesaurus, fuzzy, soundex, or about queries. Parse trees also show the following information:

• order of execution (precedence of operators)
• ABOUT query normalization
• query expression optimization
• stop-word transformations
• breakdown of composite-word tokens

Knowing how Oracle evaluates a query is useful for refining and debugging queries. You can also design your application so that it uses the explain plan information to help users write better queries.

Limitation

You cannot use EXPLAIN with remote queries.

Syntax

CTX_QUERY.EXPLAIN(


    index_name     IN VARCHAR2,
    text_query     IN VARCHAR2,
    explain_table  IN VARCHAR2,
    sharelevel     IN NUMBER DEFAULT 0,
    explain_id     IN VARCHAR2 DEFAULT NULL,
    part_name      IN VARCHAR2 DEFAULT NULL

);

index_name

Specify the name of the index to be queried.

text_query

Specify the query expression to be used as criteria for selecting rows.

When you include a wildcard, fuzzy, or soundex operator in text_query, this procedure looks at the index tables to determine the expansion.

Wildcard, fuzzy (?), and soundex (!) expression feedback does not account for lazy deletes as in regular queries.

explain_table

Specify the name of the table used to store representation of the parse tree for text_query. You must have at least INSERT and DELETE privileges on the table used to store the results from EXPLAIN.

sharelevel

Specify whether explain_table is shared by multiple EXPLAIN calls. Specify 0 for exclusive use and 1 for shared use. This parameter defaults to 0 (single-use).

When you specify 0, the system automatically truncates the result table before the next call to EXPLAIN.

When you specify 1 for shared use, this procedure does not truncate the result table. Only results with the same explain_id are updated. When no results with the same explain_id exist, new results are added to the EXPLAIN table.

explain_id

Specify a name that identifies the explain results returned by an EXPLAIN procedure when more than one EXPLAIN call uses the same shared EXPLAIN table. This parameter defaults to NULL.

part_name

Specify the name of the index partition to query.

Example

Creating the Explain Table

To create an explain table called test_explain for example, use the following SQL statement:

create table test_explain(
         explain_id varchar2(30)
         id number,
         parent_id number,
         operation varchar2(30),
         options varchar2(30),
         object_name varchar2(64),
         position number,
         cardinality number);

Executing CTX_QUERY.EXPLAIN

To obtain the expansion of a query expression such as comp% OR ?smith, use CTX_QUERY.EXPLAIN as follows:

ctx_query.explain(
         index_name => 'newindex',
         text_query => 'comp% OR ?smith',
         explain_table => 'test_explain',
         sharelevel => 0,
         explain_id => 'Test');

Retrieving Data from Explain Table

To read the explain table, you can select the columns as follows:

select explain_id, id, parent_id, operation, options, object_name, position
from test_explain order by id;

The output is ordered by ID to simulate a hierarchical query:

EXPLAIN_ID    ID PARENT_ID OPERATION    OPTIONS OBJECT_NAME POSITION 
----------- ---- --------- ------------ ------- ----------- -------- 
Test           1         0 OR           NULL    NULL          1 
Test           2         1 EQUIVALENCE  NULL    COMP%         1
Test           3         2 WORD         NULL    COMPTROLLER   1 
Test           4         2 WORD         NULL    COMPUTER      2 
Test           5         1 EQUIVALENCE  (?)     SMITH         2 
Test           6         5 WORD         NULL    SMITH         1 
Test           7         5 WORD         NULL    SMYTHE        2 

Related Topics

Chapter 3, "CONTAINS Query Operators"

Appendix H, "Stopword Transformations"

---

HFEEDBACK

In English or French, this procedure generates hierarchical query feedback information (broader term, narrower term, and related term) for the specified query.

Broader term, narrower term, and related term information is obtained from the knowledge base. However, only knowledge base terms that are also in the index are returned as query feedback information. This increases the chances that terms returned from HFEEDBACK produce hits over the currently indexed document set.

Hierarchical query feedback information is useful for suggesting other query terms to the user.

Syntax

CTX_QUERY.HFEEDBACK(
           index_name     IN VARCHAR2,
           text_query     IN VARCHAR2,
           feedback_table IN VARCHAR2,
           sharelevel     IN NUMBER DEFAULT 0,
           feedback_id    IN VARCHAR2 DEFAULT NULL,
           part_name      IN VARCHAR2 DEFAULT NULL
           );

index_name

Specify the name of the index for the text column to be queried.

text_query

Specify the query expression to be used as criteria for selecting rows.

feedback_table

Specify the name of the table used to store the feedback terms.

sharelevel

Specify whether feedback_table is shared by multiple HFEEDBACK calls. Specify 0 for exclusive use and 1 for shared use. This parameter defaults to 0 (single-use).

When you specify 0, the system automatically truncates the feedback table before the next call to HFEEDBACK.

When you specify 1 for shared use, this procedure does not truncate the feedback table. Only results with the same feedback_id are updated. When no results with the same feedback_id exist, new results are added to the feedback table.

feedback_id

Specify a value that identifies the feedback results returned by a call to HFEEDBACK when more than one HFEEDBACK call uses the same shared feedback table. This parameter defaults to NULL.

part_name

Specify the name of the index partition to query.

Example

Create HFEEDBACK Result Table

Create a result table to use with CTX_QUERY.HFEEDBACK as follows:

 
  CREATE TABLE restab ( 
    feedback_id VARCHAR2(30), 
    id          NUMBER, 
    parent_id   NUMBER, 
    operation   VARCHAR2(30), 
    options     VARCHAR2(30), 
    object_name VARCHAR2(80), 
    position    NUMBER, 
    bt_feedback ctx_feedback_type, 
    rt_feedback ctx_feedback_type, 
    nt_feedback ctx_feedback_type 
  ) NESTED TABLE bt_feedback STORE AS res_bt 
    NESTED TABLE rt_feedback STORE AS res_rt 
    NESTED TABLE nt_feedback STORE AS res_nt; 
 

CTX_FEEDBACK_TYPE is a system-defined type in the CTXSYS schema.

Call CTX_QUERY.HFEEDBACK

The following code calls the hfeedback procedure with the query computer industry.

BEGIN 
ctx_query.hfeedback (index_name     => 'my_index', 
                     text_query     => 'computer industry', 
                     feedback_table => 'restab', 
                     sharelevel     => 0, 
                     feedback_id    => 'query10' 
                    ); 
END; 

Select From the Result Table

The following code extracts the feedback data from the result table. It extracts broader term, narrower term, and related term feedback separately from the nested tables.

DECLARE 
  i NUMBER; 
BEGIN 
  FOR frec IN ( 
    SELECT object_name, bt_feedback, rt_feedback, nt_feedback  
    FROM restab 
    WHERE feedback_id = 'query10' AND object_name IS NOT NULL 
  ) LOOP 
 
    dbms_output.put_line('Broader term feedback for ' || frec.object_name || 
':'); 
    i := frec.bt_feedback.FIRST; 
    WHILE i IS NOT NULL LOOP 
      dbms_output.put_line(frec.bt_feedback(i).text); 
      i := frec.bt_feedback.NEXT(i); 
    END LOOP; 
 
    dbms_output.put_line('Related term feedback for ' || frec.object_name || 
':'); 
    i := frec.rt_feedback.FIRST; 
    WHILE i IS NOT NULL LOOP 
      dbms_output.put_line(frec.rt_feedback(i).text); 
      i := frec.rt_feedback.NEXT(i); 
    END LOOP; 
 
    dbms_output.put_line('Narrower term feedback for ' || frec.object_name || 
':'); 
    i := frec.nt_feedback.FIRST; 
    WHILE i IS NOT NULL LOOP 
      dbms_output.put_line(frec.nt_feedback(i).text); 
      i := frec.nt_feedback.NEXT(i); 
    END LOOP; 
 
  END LOOP; 
END;

Sample Output

The following output is for the example above, which queries on computer industry:

Broader term feedback for computer industry: 
hard sciences 
Related term feedback for computer industry: 
computer networking 
electronics 
knowledge 
library science 
mathematics 
optical technology 
robotics 
satellite technology 
semiconductors and superconductors 
symbolic logic 
telecommunications industry 
Narrower term feedback for computer industry: 
ABEND - abnormal end of task 
AT&T Starlans 
ATI Technologies, Incorporated 
ActivCard 
Actrade International Ltd. 
Alta Technology 
Amiga Format 
Amiga Library Services 
Amiga Shopper 
Amstrat Action 
Apple Computer, Incorporated
..

---

REMOVE_SQE

The CTX_QUERY.REMOVE_SQE procedure removes the specified stored query expression.

Syntax

CTX_QUERY.REMOVE_SQE(query_name IN VARCHAR2);

query_name

Specify the name of the stored query expression to be removed.

Examples

begin
ctx_query.remove_sqe('disasters');
end;

---

STORE_SQE

This procedure creates a stored query expression. Only the query definition is stored.

Supported Operators

Stored query expressions support all of the CONTAINS query operators. Stored query expressions also support all of the special characters and other components that can be used in a query expression, including other stored query expressions.

Privileges

Users are allowed to create and remove stored query expressions owned by them. Users are allowed to use stored query expressions owned by anyone. The CTXSYS user can create or remove stored query expressions for any user.

Syntax

CTX_QUERY.STORE_SQE(query_name      IN VARCHAR2,
                    text_query      IN VARCHAR2);

query_name

Specify the name of the stored query expression to be created. If you are CTXSYS, you can specify this as user.name.

text_query

Specify the query expression to be associated with query_name.

Examples

begin
ctx_query.store_sqe('disasters', 'hurricanes | earthquakes');
end;