| Hi, I maintain a project (diofanti.org) that tracks public spending in Greece. It’s a PG instance hosting 55M+ json documents with searching functionality on top of them. It relies heavily on to_tsvector(‘greek’, ..), as users search for company names, invoice descriptions etc. The results are fairly good, but as I was trying to experiment with adding some more domain-specific stopwords, I realised there’s no greek.stop under $(pg_config —sharedir)/tsearch_data And indeed looks like stop words are maintained with to_tsvector(‘greek’, ..). select to_tsvector('greek', 'ΚΑΛΗΜΕΡΑ ΚΑΙ ΣΕ ΕΣΑΣ'); --> 'εσ':4 'κα':2 'καλημερ':1 'σε':3 select to_tsvector('english', 'AND GOOD MORNING TO YOU TOO'); --> 'good':2 'morn’:3 I found an older discussion on pgsql-hackers [0] but not sure where this stopped / if started ? Am I missing something? Is there another thread/patch I can peek up myself ? |
