The Lemur Project Files

This Lemur project release brings Indri 5.6, Galago 3.5  and
RankLib 2.3 updates.

Applications compiled with the Indri API require the following
libraries: z, iberty, pthread, and m on linux. Applications built in
Visual Studio require the additional library wsock32.lib.  The java jar
files were built with Java 7 (jdk 1.7.0). The java UIs require Java
7. We have tested using GCC 4.1.2 (CentOS 5.3 linux), 4.3.3 (Ubuntu
10.04 linux), 4.2.1 (OS/X), and Visual Studio 2010 and 2012 (Windows
7, WIN32 and x86_64).

**Special Note**

Users should pay close attention to matching the version of Indri used to
build an index with the version used to query the Index in order to obtain
accurate search results. FR#78 added an exception if the versions are
mismatched. If you are using indexes built with a version earlier than 5.3
you must re-index your data with this version. In addition, all indexes built
with a version earlier than 5.5 that used the Krovetz stemmer must
be rebuilt.

Indri additions include:

Compilation support for VS2010 and VS2012

The ability to use a blacklist when using IndriBuildIndex


Galago additions include:

High level interface; single file index;

RankLib additions include:

Improved support cross-validation, analysis functionalities;
Coordinate Ascent now takes into account validation data;

cmd-line parameter update in the Analyzer class



Bugs Fixed (see https://sourceforge.net/p/lemur/bugs/ for the complete
tickets):

BUG#

214 -- Galago NumberedPostingsPositionExtractor null exception

215 -- Indri OffsetAnnotationAnnotator drops tokens of length 2 or less at end of field

216 -- Galago RM expansion fails on empty result set

217 -- Indri Exception running dumpindex 'c' (compact) command

218 -- Indri InferenceNetwork::_evaluateIndex assert fails when given an empty MemoryIndex

BUG -- don't throw an internal error when encountering a longrec_error.
Reported in https://sourceforge.net/p/lemur/discussion/2106523/thread/cfa7d339/

219 -- RankLib LambdaRank(with AP optimization) error

221 -- RankLib error with normalization and saved models

221 -- Indri AnchorTextAnnotator can crash on input of too many links (1M+) 



Feature Requests (see https://sourceforge.net/p/lemur/feature-requests/
for the complete tickets):

FR#

71 -- Indri Compilation support for VS2010 and VS2012

75 -- Indri Add deleteDocument method to QueryEnvironment

76 -- Indri Add a blacklist feature to IndriBuildIndex

77 -- Indri Add *stem* QueryEnvironment API calls to Java API

78 -- Indri Throw an exception when index version mismatch occurs. Update
IndriRunQuery to exit cleanly when query threads fail during initialization.