I have to do text classification with weka (SVM (SMO) algorithm)). Actually, it's a word sense disambiguation project. the files I prepared, contain some sentences (Persian language) and an English word in front of each sentence which shows its class. The files are arff. My question: for using SMO in weka for classifying these sentences, should I give weka these sentences as input or should I change the words to vector and give these vectors to weka? if I have to use the second one, according to what factor should I change words to vectors? I've enclosed a sample file.
Hi Vahideh,
In WEKA, you have to convert your texts into a word vector using the unsupervised filter "StringToWordVector". This filter *by default* makes the attribute value in the transformed dataset 1 or 0 for all single-word terms, based on whether the word appears in the document or not. After that you can invoke SMO normally.
HTH.
Samer
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification