solr_synonyms

The solr_synonyms template expands tokens through a synonyms map written in Apache Solr synonyms-file format, supplied inline via the required SYNONYMS option. It rewrites each input term to the set of terms it is equivalent to, so a search for one word also finds documents written with any of its synonyms.

Each line of the map is a rule. A comma-separated list of terms forms a bidirectional equivalence class — any term in the list expands to all of them. The arrow form lhs => rhs defines a one-way mapping — the left side rewrites to the right and never the reverse. Input that matches no rule passes through unchanged.

Use it — usually inside a pipeline after a tokenizer — to broaden recall without bloating the index, since the expansion happens at analysis time on both the indexed text and the query.

Options

Option	Type	Default	Description
SYNONYMS	string	required	Inline Solr-format synonyms: one rule per line; comma-separated terms are bidirectional, lhs => rhs maps left to right

Tokenization

A bidirectional class such as car, automobile, auto makes the three terms interchangeable: any one of them expands to all three (returned in sorted order), so a query for auto matches text that said car. A one-way rule such as laptop => notebook rewrites only in the stated direction — laptop becomes notebook, but notebook is left alone. A term that matches no rule is emitted as-is.

Input	Synonyms map	Tokens
car	car, automobile, auto	{auto,automobile,car}
automobile	car, automobile, auto	{auto,automobile,car}
laptop	laptop => notebook	{notebook}
keyboard	(no matching rule)	{keyboard}

The map below combines a bidirectional class with a one-way rule:

Query

CREATE TEXT SEARCH DICTIONARY solr_syn (    template = 'solr_synonyms',    -- comma-separated terms are bidirectional; "=>" maps left to right    synonyms = 'car, automobile, autolaptop => notebook');

Any member of the class expands to the whole class:

Query

SELECT ts_lexize('solr_syn', 'car');

Result

 ts_lexize----------------------- {auto,automobile,car}

Query

SELECT ts_lexize('solr_syn', 'automobile');

Result

 ts_lexize----------------------- {auto,automobile,car}

The left side of a one-way rule rewrites to its right side:

Query

SELECT ts_lexize('solr_syn', 'laptop');

Result

 ts_lexize------------ {notebook}

A term that matches no rule passes through unchanged:

Query

SELECT ts_lexize('solr_syn', 'keyboard');

Result

 ts_lexize------------ {keyboard}

See also

• wordnet_synonyms — WordNet-format synonyms
• pipeline — chain a tokenizer ahead of the synonym filter
• CREATE TEXT SEARCH DICTIONARY