AnalyticDB:pg_jieba

Prerequisites

Before you begin, make sure that:

• The AnalyticDB for PostgreSQL instance is in elastic storage mode

• The minor version meets the minimum requirement:

  • AnalyticDB for PostgreSQL V6.0: 6.6.2.1 or later

  • AnalyticDB for PostgreSQL V7.0: 7.0.5 or later

To check your minor version, see View the minor engine version.

Install pg_jieba

1. On the Extensions page of the AnalyticDB for PostgreSQL console, install the pg_jieba extension. For details, see Install, update, and uninstall extensions.

2. Switch to the public schema of the target database and run the following statement to verify the installation:

   SELECT * FROM pg_extension WHERE extname = 'pg_jieba';

   If the extension is installed, the output looks like this:

   +--------+--------+--------+--------+
   |oid     |extname |extowner|...     |
   +--------+--------+--------+--------+
   |17194   |pg_jieba|10.     |...     |
   +--------+--------+--------+--------+

   If no rows are returned, the extension is not installed in the public schema of that database.

Perform Chinese word segmentation

pg_jieba registers the jiebacfg text search configuration. Pass jiebacfg as the first argument to to_tsvector or to_tsquery to segment Chinese text.

Segment text into a tsvector:

SELECT to_tsvector('jiebacfg', '有两种方法进行全文检索');

Result:

+---------------------------------------+
|               to_tsvector             |
+---------------------------------------+
|'两种':2 '全文检索':5 '方法':3 '进行':4   |
+---------------------------------------+
(1 row)

Run a full-text search query:

SELECT to_tsvector('jiebacfg', '有两种方法进行全文检索') @@ to_tsquery('jiebacfg', '全文检索');

Result:

+----------+
| ?column? |
+----------+
| t        |
+----------+
(1 row)

Use a custom dictionary

By default, pg_jieba uses its built-in dictionary. For domain-specific terms—product names, technical jargon, or compound phrases—add them to the custom dictionary so they are treated as single tokens rather than split into parts.

When pg_jieba is installed, it automatically creates the jieba.jieba_custom_word table with the following schema:

CREATE TABLE jieba.jieba_custom_word
(
    word    text primary key,     -- Custom word
    weight  float8 default '1.0', -- Weight
    type    text   default 'x'    -- Part of speech
);

INSERT INTO jieba.jieba_custom_word values('两种方法');

DELETE FROM jieba.jieba_custom_word WHERE word='两种方法';

SELECT * FROM jieba.jieba_custom_word;

SELECT jieba.jieba_load_user_dict();

SELECT to_tsvector('jiebacfg', '有两种方法进行全文检索');

What's next

• Full-Text Search — PostgreSQL full-text search reference

• Text Search Functions and Operators — Reference for to_tsvector, to_tsquery, and related functions