PhoneticFilterFactory related questions

[Reuben Firmin <re...@benetech.org>]

Hi,

I have a schema with three (relevant to this question) fields: title,
author, book_content. I found that if PhoneticFilterFactory is used as a
filter on book_content, it was bringing back all kinds of unrelated results,
so I have it applied only against title and author.

Questions --

1) I have the filter set up on both the index and query analyzers for the
fieldType of title/author. When running against an index which had been
built without the phonetic filter, phonetic searches still worked. Is there
a performance benefit to applying the phonetic filter to the index analyzer
as well as the query analyzer, are there other benefits to doing so, or
should I not bother? (I.e. should I just apply the filter to the query
analyzer?)

2) Title / author matches are generally boosted, which is fine if it's an
exact match (i.e. "Shakespeare In Love" or "by William Shakespeare" are more
relevant than a book which mentions Shakespeare). However, the phonetic
filter put a bit of a spanner in the works - now if I search for "bottling",
books with the word "b*a*ttling" in the title show up above books with the
non-substituted word in the content. How can I juggle the boosting / field
setup to be something like:
a) Title/author matches (with exactly matched spelling - stemming etc is
fine)
b) Content matches (with exactly matched spelling)
c) Title/author matches (with phoneme equivalent spelling)

Do I need to create separate non-phonetic title/author fields for this, or
is there a different way to achieve the same effect?

Thanks
Reuben