lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steven Parkes" <steven_par...@esseff.org>
Subject RE: Soliciting Design Thoughts on Date Searching
Date Mon, 05 Mar 2007 20:52:48 GMT
	But, letting it stay in the text stream and not putting it in a
separate
	date
	field would give you some trouble with ranges because things
that
	weren't dates could mess you up.

This is why Chris suggested putting a prefix on the token. For example,
leading underscores: something your analyzer won't emit, so there's no
intersection between your date tokens and "normal" tokens.

I think I'm reiterating, but putting the dates in your "normal" field
allows more phrase and span queries to work. If you put the tokens into
a separate field, the positioning information won't be there.

That said, if you have a different number of tokens, e.g., December
25th, 2007 turning into 20071225, this will impact phrase and span
queries anyway so even if they are in the same field, there will be a
difference between using a date token and not.

To a first order, there should be no difference in performance, I don't
think. A posting list is a posting list, regardless of the field it
represents (again, assuming you have some prefix so normal tokens don't
end up in the middle of a date range.)

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org


Mime
View raw message