Thanks for the explanation Chris. I'll try it but the term "<lst name="queriesandreadmoreresultseventhoughthisexampleissimpleconsidercaseswheretherear"> "
strikes me as not very legitimate and the source text is just space bounded words so even if its doing what it is supposed to, I'm not sure this term is helpful in the index. I'm kinda new to TFV's though, so much to learn. On Fri, 2010-06-18 at 12:43 -0700, Chris Hostetter wrote: > : It shows some unwanted and possibly erroneous terms. > > they may be unwanted, but if it's returning them then they are in your > index ... you know the docId and field in question (it's in your URL) so > you can look at the source text, paste it into anslysis.jsp and see > exactly why those terms are being indexed based on your fieldtype -- then > change either the source data or the fieldtype analyser as needed. > > : And some improper stemming (e.g.requir? require.) > > depending on the stemmer you are using, "requir" may be a totally > legitimate root (the programatic stemmers like Porter and Snoball make no > claim that the terms they produce will be real words, just that words with > a common root will *probably* transform into the same Term) > > > > > -Hoss >
