LDA.py

"""
@Author: Arunav Mishra, Supratim Das
"""
import os
import gensim 
from gensim import similarities


class LDA(object):

    # Initialize
    def __init__(self):

        self.id2wordFile = '/Users/Supra/PycharmProjects/LDA/Input/OP/words.dict'
        self.id2word = gensim.corpora.Dictionary.load(self.id2wordFile)

        # Load the corpus files generated in the last step
        self.corpusFile = '/Users/Supra/PycharmProjects/LDA/Input/OP/corpus.mm'
        self.mm = gensim.corpora.MmCorpus(self.corpusFile)

        # Run LDA
        self.lda = gensim.models.ldamodel.LdaModel(corpus=self.mm, id2word=self.id2word, num_topics=100, update_every=0,
                                                   passes=20)
        self.ldaFile = '/Users/Supra/PycharmProjects/LDA/Input/OP/lda'
        self.dir_lda = os.path.dirname(self.ldaFile)

    # Run LDA
    def run(self):
        try:
            os.stat(self.dir_lda)
        except:
            os.mkdir(self.dir_lda)

        self.lda.save(self.ldaFile)

        index = similarities.MatrixSimilarity(self.lda[self.mm])
        dir = os.path.dirname(self.ldaFile)

        try:
            os.stat(dir)
        except:
            os.mkdir(dir)
        self.lda.save(self.ldaFile)

        ldaIndexFile = '/Users/Supra/PycharmProjects/LDA/Input/OP/lda.index'
        index.save(ldaIndexFile)

    def main(self):
        self.run()

if __name__ == '__main__':
    LDA().main()