136 lines
4.3 KiB
Python
136 lines
4.3 KiB
Python
#!/usr/bin/env python3
|
|
# -*- encoding: utf-8 -*-
|
|
|
|
# Copyright (c) 2017 Adler Neves <adlerosn@gmail.com>
|
|
#
|
|
# MIT License
|
|
#
|
|
# Permission is hereby granted, free of charge, to any person obtaining
|
|
# a copy of this software and associated documentation files (the
|
|
# "Software"), to deal in the Software without restriction, including
|
|
# without limitation the rights to use, copy, modify, merge, publish,
|
|
# distribute, sublicense, and/or sell copies of the Software, and to
|
|
# permit persons to whom the Software is furnished to do so, subject to
|
|
# the following conditions:
|
|
#
|
|
# The above copyright notice and this permission notice shall be
|
|
# included in all copies or substantial portions of the Software.
|
|
#
|
|
# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
|
# EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
|
# MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
|
|
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
|
|
# LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
|
|
# OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
|
|
# WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
|
|
|
from utils import flatten, sortedListBinarySearch, getListItemOr, dictToKeypair
|
|
|
|
with open('unitexable_train/corpus.txt.answersheet.txt') as f:
|
|
trainSet = [[[wrd.split('/',1)[0].strip().lower(),wrd.split('/',1)[1]] for wrd in snt.strip().splitlines()] for snt in f.read().strip().split('\n\n')]
|
|
|
|
print('Read from file')
|
|
|
|
tokSet = tuple(sorted(list(set(map(lambda a: a[0], flatten(trainSet))))))
|
|
tagSet = tuple(sorted(list(set(map(lambda a: a[1], flatten(trainSet))))))
|
|
|
|
print('Int conversion set')
|
|
|
|
newTrainSet = list()
|
|
for i, snt in enumerate(trainSet):
|
|
newTrainSet.append([
|
|
(sortedListBinarySearch(tokSet,wrd[0]),sortedListBinarySearch(tagSet,wrd[1]))
|
|
for wrd in snt
|
|
])
|
|
if(i%10000 == 0):
|
|
print('%d of %d'%(i,len(trainSet)))
|
|
|
|
print('Int conversion sentences')
|
|
|
|
def makeEmptyTriples(lst):
|
|
triples = list()
|
|
for i in range(1,len(lst)-1):
|
|
triples.append(((lst[i-1][0], lst[i+1][0]),lst[i][1]))
|
|
return triples
|
|
|
|
def makeTriples(lst):
|
|
triples = list()
|
|
for i in range(len(lst)):
|
|
triples.append(((getListItemOr(lst, i-1, [None])[0], lst[i][0], getListItemOr(lst, i+1, [None])[0]),lst[i][1]))
|
|
return triples
|
|
|
|
def makeTuplesL(lst):
|
|
tuples = list()
|
|
for i in range(1, len(lst)):
|
|
tuples.append(((lst[i-1][0], lst[i][0]),lst[i][1]))
|
|
return tuples
|
|
|
|
def makeTuplesR(lst):
|
|
tuples = list()
|
|
for i in range(len(lst)-1):
|
|
tuples.append(((lst[i][0], lst[i+1][0]),lst[i][1]))
|
|
return tuples
|
|
|
|
def makeFallback(lst):
|
|
fallback = list()
|
|
for i in range(len(lst)-1):
|
|
fallback.append(((lst[i][0],),lst[i][1]))
|
|
return fallback
|
|
|
|
tgfb = tuple(flatten([makeFallback(snt) for snt in newTrainSet]))
|
|
tgtl = tuple(flatten([makeTuplesL(snt) for snt in newTrainSet]))
|
|
tgtr = tuple(flatten([makeTuplesR(snt) for snt in newTrainSet]))
|
|
tgtp = tuple(flatten([makeTriples(snt) for snt in newTrainSet]))
|
|
tget = tuple(flatten([makeEmptyTriples(snt) for snt in newTrainSet]))
|
|
|
|
print('tuples created')
|
|
|
|
def makeFrequencyTupleDict(tuples):
|
|
d = dict()
|
|
for t in tuples:
|
|
if t[0] not in d:
|
|
d[t[0]] = dict()
|
|
if t[1] not in d[t[0]]:
|
|
d[t[0]][t[1]] = 1
|
|
else:
|
|
d[t[0]][t[1]]+= 1
|
|
return d
|
|
|
|
def getMostFrequentTupleDict(d1):
|
|
d = dict()
|
|
for tpl,freq in d1.items():
|
|
d[tpl] = sorted([tuple([freq, tag]) for tag,freq in freq.items()])[-1][1]
|
|
return d
|
|
|
|
kpfb = dictToKeypair(getMostFrequentTupleDict(makeFrequencyTupleDict(tgfb)))
|
|
kptl = dictToKeypair(getMostFrequentTupleDict(makeFrequencyTupleDict(tgtl)))
|
|
kptr = dictToKeypair(getMostFrequentTupleDict(makeFrequencyTupleDict(tgtr)))
|
|
kptp = dictToKeypair(getMostFrequentTupleDict(makeFrequencyTupleDict(tgtp)))
|
|
kpet = dictToKeypair(getMostFrequentTupleDict(makeFrequencyTupleDict(tget)))
|
|
|
|
print('decision keypairs created')
|
|
|
|
import json
|
|
|
|
sv = {
|
|
'tagset':tagSet,
|
|
'tokset':tokSet,
|
|
'keypairs': {
|
|
'1fallback': kpfb,
|
|
'2left': kptl,
|
|
'2right': kptr,
|
|
'3middle':kptp,
|
|
'3empty': kpet
|
|
}
|
|
}
|
|
|
|
svt = json.dumps(sv)
|
|
|
|
print('serialized train data')
|
|
|
|
with open('unitexable_test/traindata.json','w') as f:
|
|
f.write(svt)
|
|
|
|
print('saved train data in disk')
|
|
print('done')
|