varvara-l

import sys

'''
Parse TERCOM .pra files
'''

def parse_sentence(line_array):
    hyp, ref = [], []
    align, sentence_id = "", ""
    for line in line_array:

varvara-l

from __future__ import print_function
from argparse import ArgumentParser
from sklearn.metrics import accuracy_score
###########################################################################################################
#
#   Script to compute Sequence Correlation score for word-level binary QE system output in WMT-15 format: 
#               <METHOD NAME> <SEGMENT NUMBER> <WORD INDEX> <WORD> <BINARY SCORE> 
#
###########################################################################################################

varvara-l

from __future__ import division

import sys
import numpy as np
import math
from argparse import ArgumentParser
from multiprocessing import Pool

##########################################################################
#

varvara-l

from __future__ import division

import sys
import numpy as np
import math
from argparse import ArgumentParser
from multiprocessing import Pool

##########################################################################
#

varvara-l

from __future__ import division
from argparse import ArgumentParser
import sys
from scipy.stats import pearsonr, t
from math import sqrt

########################################################################################
#  Williams significance test to define significance of Pearson correlation scores
#  Implemented as described in [1]. The final significance level is modified
#  with Bonferroni correction for multiple comparisons

varvara-l

from __future__ import division, print_function

import os
import numpy as np
from argparse import ArgumentParser
from sklearn.metrics import f1_score
#------------------ evaluation for the WMT15 format:------------------------
#
#       <METHOD NAME> <SEGMENT NUMBER> <WORD INDEX> <WORD> <BINARY SCORE>
#       tab-separated, no empty lines