Elasticsearch analysis plugin

this plugin is used to join every possible two-word combinations (pairs) in a document.

the plugin exports a token filter called wordjoin_stemmer.

basically this filter stems every token from the stream and joins the original word and the stemmed one with every subsequent word in the document. this is useful in e-commerce websites where documents are small and a user might search with two words joined as a single word.

Stemming happens on two steps

• the first is using stanfordnlp/CoreNLP Stemmer class with small modifications.
• the second step is applying custom stemming rules in the CustomStemmer class.

to use the plugin please follow these steps:

• clone the repo on your local machine
• make sure you have maven installed on the machine
• mvn -Dtest=EsAnalysisTests test. this will run tests to make sure everything is ok. you should always run tests if you change anything in the plugin
• mvn package. this will package the project and output a zip file under target/releases folder in the project directory
• sudo /usr/share/elasticsearch/bin/elasticsearch-plugin install file:///home/ubuntu/es-analyzer-plugin-7.0.0.zip install the plugin in elasticsearch using elasticsearch-plugin executable installed with elasticsearch. the path of this will vary depending on the way you installed elasic on the machine.
• if the plugin is already installed you will need to remove it first.