Home

Awesome

Vietnamese Analysis Plugin for Elasticsearch

Test

Vietnamese Analysis plugin integrates Vietnamese language analysis into Elasticsearch. It uses C++ tokenizer for Vietnamese library developed by CocCoc team for their Search Engine and Ads systems.

The plugin provides vi_analyzer analyzer, vi_tokenizer tokenizer and vi_stop stop filter. The vi_analyzer is composed of the vi_tokenizer tokenizer, stop and lowercase filter.

Example output

GET _analyze
{
  "analyzer": "vi_analyzer",
  "text": "Cộng hòa Xã hội chủ nghĩa Việt Nam"
}

The above sentence would produce the following terms:

["cộng hòa", "xã hội", "chủ nghĩa" ,"việt nam"]

Configuration

The vi_analyzer analyzer accepts the following parameters:

Example configuration

In this example, we configure the vi_analyzer analyzer to keep punctuation marks and to use the custom list of stop words:

PUT my-vi-index-00001
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_vi_analyzer": {
          "type": "vi_analyzer",
          "keep_punctuation": true,
          "stopwords": ["rất", "những"]
        }
      }
    }
  }
}

GET my-vi-index-00001/_analyze
{
  "analyzer": "my_vi_analyzer",
  "text": "Công nghệ thông tin Việt Nam rất phát triển trong những năm gần đây."
}

The above example produces the following terms:

["công nghệ", "thông tin", "việt nam", "phát triển", "trong", "năm", "gần đây", "."]

We can also create a custom analyzer with the vi_tokenizer. In following example, we create my_vi_analyzer to produce both diacritic and no diacritic tokens in lowercase:

PUT my-vi-index-00002
{
  "settings": {
    "analysis": {
      "analyzer": {
        "my_vi_analyzer": {
          "tokenizer": "vi_tokenizer",
          "filter": [
            "lowercase",
            "ascii_folding"
          ]
        }
      },
      "filter": {
        "ascii_folding": {
          "type": "asciifolding",
          "preserve_original": true
        }
      }
    }
  }
}

GET my-vi-index-00002/_analyze
{
  "analyzer": "my_vi_analyzer",
  "text": "Cộng hòa Xã hội chủ nghĩa Việt Nam"
}

The above example produces the following terms:

["cong hoa", "cộng hòa", "xa hoi", "xã hội", "chu nghia", "chủ nghĩa", "viet nam", "việt nam"]

Use Docker

Make sure you have installed both Docker & docker-compose

Build the image with Docker Compose

# Copy, edit ES version and password for user elastic in file .env. Default password: changeme
cp .env.sample .env
docker compose build
docker compose up

Verify

curl -k http://elastic:changeme@localhost:9200/_analyze -H 'Content-Type: application/json' -d '
{
  "analyzer": "vi_analyzer",
  "text": "Cộng hòa Xã hội chủ nghĩa Việt Nam"
}'

# Output
{"tokens":[{"token":"cộng hòa","start_offset":0,"end_offset":8,"type":"<WORD>","position":0},{"token":"xã hội","start_offset":9,"end_offset":15,"type":"<WORD>","position":1},{"token":"chủ nghĩa","start_offset":16,"end_offset":25,"type":"<WORD>","position":2},{"token":"việt nam","start_offset":26,"end_offset":34,"type":"<WORD>","position":3}]}                                                                                     

Build from Source

Step 1: Build C++ tokenizer for Vietnamese library

git clone https://github.com/duydo/coccoc-tokenizer.git
cd coccoc-tokenizer && mkdir build && cd build
cmake -DBUILD_JAVA=1 ..
make install
# Link the coccoc shared lib to /usr/lib
sudo ln -sf /usr/local/lib/libcoccoc_tokenizer_jni.* /usr/lib/

By default, the make install installs:

Verify

/usr/local/bin/tokenizer "Cộng hòa Xã hội chủ nghĩa Việt Nam"
# cộng hòa	xã hội	chủ nghĩa	việt nam

Refer the repo for more information to build the library.

Step 2: Build the plugin

Clone the plugin’s source code:

git clone https://github.com/duydo/elasticsearch-analysis-vietnamese.git

Optionally, edit the elasticsearch-analysis-vietnamese/pom.xml to change the version of Elasticsearch (same as plugin version) you want to build the plugin with:

...
<version>8.7.0</version>
...

Build the plugin:

cd elasticsearch-analysis-vietnamese
mvn package

Step 3: Installation the plugin on Elasticsearch

bin/elasticsearch-plugin install file://target/releases/elasticsearch-analysis-vietnamese-8.7.0.zip

Compatible Versions

From v7.12.11, the plugin uses CocCoc C++ tokenizer instead of the VnTokenizer by Lê Hồng Phương, I don't maintain the plugin with the VnTokenizer anymore, if you want to continue developing with it, refer the branch vntokenizer.

Vietnamese Analysis PluginElasticsearch
master8.7.0
develop8.7.0
8.7.08.7.0
8.4.08.4.0 ~ 8.7.1
8.0.08.0.0 ~ 8.0.x
7.16.17.16 ~ 7.17.1
7.12.17.12.1 ~ 7.15.x
7.3.17.3.1
5.6.55.6.5
5.4.15.4.1
5.3.15.3.1
5.2.15.2.1
2.4.12.4.1
2.4.02.4.0
2.3.52.3.5
2.3.42.3.4
2.3.32.3.3
2.3.22.3.2
2.3.12.3.1
2.3.02.3.0
0.2.22.2.0
0.2.1.12.1.1
0.2.12.1.0
0.22.0.0
0.1.71.7+
0.1.61.6+
0.1.51.5+
0.1.11.4+
0.11.3

Issues

You might get errors during starting Elasticsearch with the plugin

1. Error: java.lang.UnsatisfiedLinkError: no libcoccoc_tokenizer_jni in java.library.path ... (reported in 102)

It happens because of your JVM cannot find the dynamic lib libcoccoc_tokenizer_jni in java.library.path, try to resolve by doing one of following options:

export LD_LIBRARY_PATH=/usr/local/lib:$LD_LIBRARY_PATH
# Make link
ln -sf /usr/local/lib/libcoccoc_tokenizer_jni.so /usr/lib/libcoccoc_tokenizer_jni.so

# Copy 
cp /usr/local/lib/libcoccoc_tokenizer_jni.so /usr/lib

2. Error: Cannot initialize Tokenizer: /usr/local/share/tokenizer/dicts (reported in 106)

It happens because of the tokenizer cannot find the dictionary files under /usr/local/share/tokenizer/dicts. Ensure the path /usr/local/share/tokenizer/dicts existed and includes those files: alphabetic, i_and_y.txt, nontone_pair_freq_map.dump, syllable_trie.dump d_and_gi.txt, multiterm_trie.dump, numeric. If not, try to build the C++ tokenizer (Step 1) again.

Thanks to

License

This software is licensed under the Apache 2 license, quoted below.

Copyright by Duy Do

Licensed under the Apache License, Version 2.0 (the "License"); you may not
use this file except in compliance with the License. You may obtain a copy of
the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS, WITHOUT
WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the
License for the specific language governing permissions and limitations under
the License.