A package for large and small scale indexing

This package ( se.illuminet.index ) handles indexing of more or less structured information.

There are two types of indexes:

a quick store memory managed (foward) barrel (in the mem package)
a quick read (inverted) storage for large indexes on disk

Booth are persistant to disk (on sync) and contain tokens related to:

Page
Context (structural or/and lingvistic reference)
Offset

Tracking contexts

Contexts are stored in a tree-structure all related to Tag's

Indexing

The "Indexer" is used with the content package for indexing. By default Indexer uses one mem barrel and one fat (disk) barrel for index-storage.

Goal

We have a generic indexing API for serveral types of indexes. This index package relies on dictionaries to supply tokens as integers and store handles about each reference list.

Developer

Jonas Bosson (jonas at illuminet dot se)

News

2004-05-06: Made code LGPL

Subprojects

IndexService
mem & fwd (forward fast indexing)
fat (reverse indexes in store)

License: Open Source

This software is published under LGPL-license (Lesser GNU Public License)

Requires

Java 1.4 or higher (or compatible)
se.illuminet.store
optional:

babel
corpus

Download

http://www.illuminet.se/software/build/

CVS

http://www.illuminet.se/cgi-bin/cvsweb.cgi/public/src/se/illuminet/index/