The AutoPhrase TokenFilter indexes multi-word expressions as single tokens, so New York City is one thing rather than three. It is not installed on every Opensolr environment, but we will put it on yours.
01 · How to tell whether it is active
If the filter is missing, your index will refuse to load with an error naming the factory class:
Plugin not found: solr.AutoPhraseTokenFilterFactory
02 · Getting it installed
Send us the request
Contact us and we will install the AutoPhrase library, or any other custom Solr plugin, on your index.
Follow the jar procedure
The general process, and what to include, is in how to add a lib jar file.
Tell us your Solr version
The jar has to match the major version your index runs, so name it in the request.
03 · After it is installed
Add the filter to your field type
Put the <filter class="solr.AutoPhraseTokenFilterFactory" ... /> element into the analysis chain of the field type that needs it, in schema.xml.
Reload the index
The filter is active after a reload from the Tools menu.
Provide its phrase list
AutoPhrase needs its own configuration, including the list of phrases it should recognise. Upload it with your other configuration files.
04 · Worth knowing first
Version compatibility is not optional
A filter built against another Solr version will stop the core from loading, not just the field.
Every jar is reviewed
Public and official plugins are approved faster, because their provenance is verifiable.
It costs indexing time
Heavy analysis chains slow indexing down. Test with your real documents, not with a sample of ten.
05 · References
The plugin itself, with its configuration options.
Solr reference guide: token filters
How a filter fits into an analysis chain, with every built-in filter Solr ships.