SVMlight

1
SVMlight

• Studies in Computational Linguistics IIOpinion
  Mining and Sentiment Analysis
• Hwang Inbeom

2
Installation

• Windows
• Windows binary
• http//download.joachims.org/svm_light/current/svm
  _light_windows.zip
• Unpack this, and run
• It may be needed an additional file,
• jvm.dll
• http//www.dlldump.com/cgi-bin/testwrap/downloadco
  unts.cgi?rtcountpathdllfiles/J/jvm.dll

3
Installation (contd.)

• Linux
• Binary file was not runnable under Fedora Core 9
  on IBM Power PC
• Download the source code and compile it
• http//download.joachims.org/svm_light/current/svm
  _light.tar.gz
• Cygwin
• May be in the same way with the Linux case

4
svm_learn

• Training phase
• svm_learn example1/train.dat example1/model

5
svm_learn (contd.)

• Data file format
• 1 60.0198403253586671 150.0339873732306071
  290.0360280968798065 310.0378103484117687

6
svm_learn

• svm_learn generates a model file, which is in
  this case example1/model
• SVM-light Version V6.010 kernel type3
  kernel parameter -d 1 kernel parameter -g 1
  kernel parameter -s 1 kernel parameter -r
  empty kernel parameter -u 9947 highest
  feature index 2000 number of training
  documents 879 number of support vectors plus 1
  -0.13217617 threshold b, each following line
  is a SV (starting with alphay)

7
svm_classify

• Testing phase
• svm_classify example1/test.dat example1/model
  example1/out

8
svm_classify

• Data file is in the same format with training set
• Output file
• Each line contains the value of the decision
  function on each element in test set