Text Classification with SVM

Happy New Year. Thank you for your continued support this year. I tried out SVM (Support Vector Machine), so I’ll write down the method. Like Naive Bayes, it can classify text. Here, I used svm_light.

cd  /usr/local/src
mkdir svm_light
cd svm_light
wget http://download.joachims.org/svm_light/current/svm_light_linux.tar.gz
gunzip -c svm_light.tar.gz | tar xvf -
make all

After installation, add /usr/local/src/svm_light to your PATH.

1 i am happy
1 happy happy new year
-1 bad year
-1 i am bad

Next, create a correspondence table between each word and a number.

1:i
2:am
3:happy
4:new
5:year
6:bad

Finally, create training data based on this correspondence table (call this file “example_file”).

1  1:1 2:1 3:1
1  3:2 4:1 5:1
-1 5:1 6:1
-1 1:1 2:1 6:1

svm_learn example_file model_file

Create a model file (“model_file”) with the svm_learn command.