SpeechAct

code used to prepare Switchboard Dialog Act corpus in the form of vector for SVM classifier

preprocess.py

* read the SWDA corpus 
* uses Porter Stemmer to convert words to there stem 
* stop words which found in "corpus/stopword" will be removed  
* words with frequency count less than 10 is removed 

dictionary.py

* prepare a dictionary using gensim library and the preprocessed data 

DocToBOw.py

* create a BOW corpus using the dictionary created before 

lsi_topic_trans.py/lda_topic_trans.py

* transform the BOW corpus to other representation using LSI/LDA 

postprocess.py

* append class label to the vectors 

final-cleanup.py

* remove unnecessary characters and prepare the final corpus for LIBSVM