Question

For example, I have a table which contains the columns:

english_description
french_description
italian_description

When using ThinkingSphinx, I can specify what stemmer to use in config/sphinx.yml like this:

development:
  morphology: stem_en

but this will run all of the columns through the english stemmer. Is there a way to specify the stemmer for each column? Something like:

development:
  morphology:
    english_description: stem_en
    french_description: libstemmer_fr
    italian_description: libstemmer_it
    ...

Or any other way to accomplish the same?

Answer 1

Morphologies/stemmers are an index-based setting from Sphinx’s perspective – so you’ll need to have separate indices for each language. With each one, make sure you have the same attributes, otherwise Sphinx will complain (or at least be less reliable). Note the explicit index names below:

define_index('model_en') do
  indexes english_description

  set_property :morphology => 'stem_en'
end

define_index('model_fr') do
  indexes french_description

  set_property :morphology => 'libstemmer_fr'
end

# etc