Hmmm, I'm nowhere near an expert on how the analyzers actually work, so I have to punt a bit here. And certainly take any of "the regulars" advice if they give it <G>...
But outside of stemming, Lucene/SOLR really doesn't understand the concept of "language". And that's not even Lucene, it's the stemmer code. The Analyzers are just concerned with producing tokens. There are some special cases where, say, accents are folded. Various European languages have accent, grave and unaccented characters for instance, which should all be treated as one character for a good search experience. See IsoLatin1AccentFilter. But as I remember (OK, it's 35 years ago that I had 2 years of Russian in college, OK?) the cyrillic alphabet doesn't suffer from that kind of problem, so it's probably worth giving it a try. At very worst, you could pre-process your indexed text and query text to smooth out any anomalies. If you want to dig farther, you could make your own analyzer..... HTH Erick On Mon, Mar 1, 2010 at 4:31 PM, michaelnazaruk <michaelnaza...@gmail.com>wrote: > > Thank you! And one little question: > Can I use RussianAnalyzer for ukrainian characters? > -- > View this message in context: > http://old.nabble.com/Cyrillic-problem-tp27744106p27749323.html > Sent from the Solr - User mailing list archive at Nabble.com. > >
