dbpedia2vec.py 文件源码

def main():
  parser = argparse.ArgumentParser(prog="DBpedia2vec", 
                                    description="DBpedia2vec embeddings training using Wikipedia dump",
                                    formatter_class=argparse.RawTextHelpFormatter)
  parser.add_argument('--i', metavar='wikipedia xml input file', dest="INFILE", type=str, default=sys.stdin, help="<input-file>", required=True,)
  args = parser.parse_args()


  #STEP 1 : parsing wiki documents from xml
  a=WikiXmlHandler()
  a.parse(args.INFILE)
  print "Step 1 complete"

  #STEP 2 : reading wkidocuments and joining the page link tokens with underscore into a single token
  x=PagesToSentences()
  x.processWikiPagesData(a.pages_file)
  print "Step 2 complete"

  #STEP 3 : word2vec training of sentences
  w2v=Word2VecTraining()
  w2v.train(x.sentencefilename,"../output/model","../output/vocab.txt")  
  print "Step 3 complete. Check output folder!!"