java-user@lucene.apache.org
[Top] [All Lists]

Search algo for the postings ( or TermFreqs)

Subject: Search algo for the postings or TermFreqs
From: Prasenjit Mukherjee
Date: Tue, 25 Apr 2006 21:55:35 +0530
Given a term "myterm", what kind of search algorithm lucene uses to get to the postings list(i.e. the term-frequency location in .frq file) ? From what I understood by looking into the lucene fileformat, is that it keeps the whole of .tii file in memory and and does a skipped linear search of the TermInfoIndex(present in .tii file) for the term "myterm".

I have gone through Doug's article titled "Space Optimizations for Total Ranking", but that seems to talk mainly about efficiently calculating the score by parallely merging the different posting streams. I didnt find any reference in that article on how to get to the postings, or may be I overlooked something there ?

Any pointers will be greatly appreciated.

thanks,
prasen

---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@xxxxxxxxxxxxxxxxx
For additional commands, e-mail: java-user-help@xxxxxxxxxxxxxxxxx

<Prev in Thread] Current Thread [Next in Thread>
  • Search algo for the postings ( or TermFreqs), Prasenjit Mukherjee <=