def run(dev, processed): file_name = "RTE2_dev.preprocessed.xml" data_set = part1a.run() texts = data_set[0] hypos = data_set[1] parta = "wordmatches.txt" partb_1 = "lemma_matches.txt" partb_2 = "pos-tag_matches.txt" partc = "bleuresults.txt" partd = "idfresults.txt" partd_threshold = 0.9670 partc_threshold = 0.2690 parta_threshold = 0.8190 partb_1_threshold = 0.9170 partb_2_threshold = 0.5790 part1b.run(file_name) part1c.run(texts, hypos) part1d.predict(texts, hypos) make_predictions.predict(parta_threshold, parta) make_predictions.predict(partb_1_threshold, partb_1) make_predictions.predict(partb_2_threshold, partb_2) make_predictions.predict(partc_threshold, partc) make_predictions.predict(partd_threshold, partd)
def run(dev, processed):
#***********************************************************
# Extracts the text enclosed in <t></t>
# Returns a list of characters
#***********************************************************
class TextHandler(ContentHandler):
in_text = False
def __init__(self, text):
ContentHandler.__init__(self)
self.text = text
self.data = []
def startElement(self, name, attrs):
if name == 't':
self.in_text = True
def endElement(self, name):
if name == 't':
t = ''.join(self.data)
self.data = []
self.text.append(t)
self.in_text = False
def characters(self, string):
if self.in_text:
self.data.append(string)
#***********************************************************
# Extracts the text enclosed in <h></h>
# Returns a list of characters
#***********************************************************
class HypothesisHandler(ContentHandler):
in_hypothesis = False
def __init__(self, hypothesis):
ContentHandler.__init__(self)
self.hypothesis = hypothesis
self.data = []
def startElement(self, name, attrs):
if name == 'h':
self.in_hypothesis = True
def endElement(self, name):
if name == 'h':
t = ''.join(self.data)
self.data = []
self.hypothesis.append(t)
self.in_hypothesis = False
def characters(self, string):
if self.in_hypothesis:
self.data.append(string)
# Uses the content handlers to extract texts and hypothesis from the xml-file
text = []
hypothesis = []
parse(dev, TextHandler(text))
parse(dev, HypothesisHandler(hypothesis))
# converts everything to lowercase
text = map(lambda x: x.lower(), text)
hypothesis = map(lambda x: x.lower(), hypothesis)
no_words = []
texts = []
hypos = []
# TODO: creating lists of words and removal of punctuation can be done on the text-list as awhole
# instead of for each element
for i in range(len(text)):
t = text[i]
h = hypothesis[i]
# create lists of words from the lists of characters
t = t.split()
h = h.split()
# remove punctuations TODO: Extend the list of characters to be removed
t = map(lambda x: x.strip('.,:;()"'), t)
h = map(lambda x: x.strip('.,:;()"'), h)
texts.append(t)
hypos.append(h)
bleu = "bleuresults.txt"
idf = "idfresults.txt"
step_size = 0.001
part1d.predict(texts, hypos)
part1c.run(text, hypos)
predict.predict(step_size, bleu)
predict.predict(step_size, idf)