Here's my understanding of the current state of analyzer, which might be wrong.
The analysis-xx plugin provides the stemming function to the analyzer used for indexing but it does not provide the same stemming function the the query analyzer. This means that instead of typing the complete word in query, you'd have to type the stem of the word. This is like typing in "chang" instead of "change", if stemming were done in English, since "chang" is the stem of the word "change" and its inflectional forms. The right thing to do is to make the analyzer used for query plugable. Challenge here is that there is no way to detect the language of query, because the query strings are too short for that. The language must be provided by query application in an application specific way such as asking the user a language by drop-down list etc. -kuro > -----Original Message----- > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] > Sent: 2006-6-18 3:33 > To: [email protected] > Subject: Re: stemming > > Is there any way to setup stemming? I made necessary changes in > includes by adding analysis-(de|fr|ru), but it seems to me that there > is a problem with "search query module", because the words that > present in index the nutch couldn't find (even with the right form of > the words). What is my mistake? > > _______________________________________________ Nutch-general mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/nutch-general
