liblinear-ruby-swig

DESCRIPTION:

This is the Ruby LIBLINEAR SWIG (Simplified Wrapper and Interface Generator) interface. LIBLINEAR is a high performance machine learning library for large scale text mining(www.csie.ntu.edu.tw/~cjlin/liblinear).

A slightly modified version of LIBLINEAR 1.51 is included which allows turning on/off the default debuging/logging messages. You don’t need your own copy of SWIG to use this library - all needed files are generated using SWIG already.

The Ruby LIBLINEAR interface was used to build a decent sized text classification model(with 10 classes and 10000 features) using 100,000 text records (through ActiveRecord in a Rails app) as training data, the recall rate is 99% and predication rate is around 90% on 2000 test records. The same model was tried with arbitrary and web page texts with pretty good results.

LIBLINEAR is in use at tweetsentiments.com - A Twitter / Tweet sentiment analysis application

INSTALL:

Currently the gem is available on linux only(tested on Ubuntu 8-9 and Fedora 9-12, and should work on OS X), and you will need g++ installed to compile the native code. 

sudo gem sources -a http://gems.github.com   (you only have to do this once)
sudo gem install tomz-liblinear-ruby-swig

SYNOPSIS:

Try the following multiclass problem in irb:

irb(main):001:0> require 'rubygems'
irb(main):002:0> require 'linear'
irb(main):003:0> pa = LParameter.new
irb(main):004:0> pa.solver_type = L2LOSS_SVM_DUAL
irb(main):005:0> pa.eps = 0.1
irb(main):006:0> bias = 1
irb(main):007:0> labels = [1, 2, 1, 2, 3]
irb(main):008:0> samples = [
irb(main):009:1*            {1=>0,2=>0.1,3=>0.2,4=>0,5=>0},
irb(main):010:1*            {1=>0,2=>0.1,3=>0.3,4=>-1.2,5=>0},
irb(main):011:1*            {1=>0.4,2=>0,3=>0,4=>0,5=>0},
irb(main):012:1*            {1=>0,2=>0.1,3=>0,4=>1.4,5=>0.5},
irb(main):013:1*            {1=>-0.1,2=>-0.2,3=>0.1,4=>1.1,5=>0.1}
irb(main):014:1>           ]
irb(main):015:0> pa.solver_type = MCSVM_CS
irb(main):016:0> sp = LProblem.new(labels,samples,bias)
irb(main):017:0> m = LModel.new(sp, pa)
irb(main):018:0>  pred = m.predict({1=>1,2=>0.1,3=>0.2,4=>0,5=>0})
=> 1
irb(main):019:0>  pred = m.predict({1=>0,2=>0.1,3=>0.2,4=>0,5=>0})
=> 2
irb(main):020:0>  pred = m.predict({1=>0,2=>0.1,3=>0.2,4=>0,5=>0})
=> 2
irb(main):025:0>  pred = m.predict({1=>0.4,2=>0,3=>0,4=>0,5=>0})
=> 1
irb(main):021:0>  pred = m.predict({1=>-0.1,2=>-0.2,3=>0.1,4=>1.1,5=>0.1})
=> 3
irb(main):022:0> m.save("test.model")
irb(main):023:0> "your imagination " * 10

For more examples see test*.rb in the liblinear-ruby-swig/liblinear-1.51/ruby directory

AUTHOR:

Tom Zeng