Remove MemoryError issue

[silvioOlivastri]

Fix MemoryError: #13

[silvioOlivastri]

If you want I have new FAST version:

def getWordmapFast(wordfile):
    wordEmbedding = pd.read_table(wordfile, sep=" ", index_col=0, header=None, quoting=csv.QUOTE_NONE)
    words = {}
    We = np.zeros(shape=(wordEmbedding.shape[0], wordEmbedding.shape[1]))
    n = 0
    for row in wordEmbedding.itertuples():
        words[row[0]] = n
        We[n] = row[1:]
        n+=1
    return (words, We)

The bad thing is the approximation. I don't know why, but few values of glove are little bit different when into wordEmbedding variables.