BagOfWordsData.py 文件源码

def _calcWordTypeCooccurMatrix(self, Q, sameWordVec, nDoc):
        """ Transform building blocks into the final Q matrix

        Returns
        -------
        Q : 2D array, size W x W (where W is vocab_size)
        """
        Q /= np.float32(nDoc)
        sameWordVec /= np.float32(nDoc)
        diagIDs = np.diag_indices(self.vocab_size)
        Q[diagIDs] -= sameWordVec

        # Fix small numerical issues (like diag entries of -1e-15 instead of 0)
        np.maximum(Q, 1e-100, out=Q)
        return Q