this is based on calsyslab project
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 

66 lines
2.0 KiB

import MarketPlaces.Classifier.transformer
import pickle, re
class Transformer:
def __init__(self):
self.ngram_list = ['res', 'vir', 'att', 'tta', 'tac', 'web', 'cat', 'rse', 'cra', 'rac', 'nso', 'omw', 'mwa', 'tec', 'boo', 'adv', 'abl', 'can', 'mmi', 'cke', 'bot', 'oks', 'ick', 'eak', 'whe', 'val', 'acc', 'mon', 'dvi', 'nto', 'phi', 'deo', 'hao', 'aos', 'pst', 'ddo', 'dos', 'iru', 'kit', 'jac']
self.ngram_index_dict = self.get_ngram_index()
def get_ngram_index(self):
ngram_index = {}
index = 0
for ngram in self.ngram_list:
if ngram not in ngram_index:
ngram_index[ngram] = index
index += 1
return ngram_index
def binary_vector(self, text):
vec = [0] * len(self.ngram_index_dict)
for ngram, index in self.ngram_index_dict.items():
if ngram in text:
vec[index] = 1
return vec
def frequency_vector(self, text):
vec = [0] * len(self.ngram_index_dict)
for ngram, index in self.ngram_index_dict.items():
vec[index] = text.count(ngram)
return vec
def relative_frequency_vector(self, freq_vec):
total = 0
for count in freq_vec:
total += count
vec = [0] * len(freq_vec)
if total > 0:
for i, count in enumerate(freq_vec):
vec[i] = count / total
return vec
def transform(self, text):
clean_sent = re.sub(r'[^a-zA-Z ]', '', text).lower()
bin_vec = self.binary_vector(clean_sent)
return bin_vec
def load(file_path):
with open(file_path, 'rb') as f:
return pickle.load(f)
def fix(fname):
t = load(fname)
new_t = MarketPlaces.Classifier.transformer.Transformer()
new_t.ngram_list = t.ngram_list
new_t.ngram_index_dict = t.ngram_index_dict
MarketPlaces.Classifier.transformer.save(new_t, fname + ".new")
# fix("desc_transformer.pickle")
fix("product_title_transformer.pickle")