def pretreate() :
from pretreate.run import Corpus
from file.path_manager import PathManager
articles_path = PathManager.CORPUS_ARTICLE
participle_title_path = PathManager.CORPUS_SPLIT
treated_article_path = PathManager.CORPUS_FEATURE
corpus = Corpus()
corpus.run(articles_path, participle_title_path, treated_article_path)
def embedding() :
from embedding.run import Corpus
from file.path_manager import PathManager
articles_path = PathManager.CORPUS_ARTICLE
participle_title_path = PathManager.CORPUS_SPLIT
sentences_path = PathManager.CORPUS_KEYWORD
word_embedding_path =PathManager.CORPUS_SIMPLYARTICLE
corpus = Corpus()
corpus.run(articles_path, participle_title_path, sentences_path, sentences_path, \
word_embedding_path, word_embedding_path)
def tag() :
from tag.run import Corpus
from file.path_manager import PathManager
sentences_path = PathManager.CORPUS_ARTICLE
tag_tree_path = PathManager.TAG_TAGTREE
tags_path = PathManager.TAG_ARTICLETAG
tags_market_path = PathManager.TAG_ARTICLETAG
sentences_market_path = PathManager.TAG_SENTENCES
untag_sentence_path = PathManager.TAG_UNTAGSENTENCE
corpus = Corpus()
corpus.run(sentences_path, tag_tree_path, sentences_market_path, tags_path, \
tags_market_path, untag_sentence_path)
print 'finish'
def bowlr() :
from bowlr.run import Corpus
from file.path_manager import PathManager
articles_path='E:/data/knowledge/bowlr/car/article_trainset'
article_market_path='E:/data/knowledge/bowlr/car/article_trainset_market'
dictionary_path='E:/data/knowledge/bowlr/all/dictionary'
feature_path='E:/data/knowledge/bowlr/car/trainset_feature'
feature_market_path='E:/data/knowledge/bowlr/car/trainset_feature_market'
train_path='E:/data/knowledge/bowlr/car/trainset_feature_market'
test_path='E:/data/knowledge/bowlr/car/testset_feature_market'
output_path='E:/data/knowledge/bowlr/car/fprs_tprs'
corpus = Corpus()
corpus.run(articles_path, article_market_path, dictionary_path, feature_path, \
feature_market_path, train_path, test_path, output_path)
def classify() :
from classify.run import Corpus
from file.path_manager import PathManager
articles_path='E:/data/knowledge/classify/car/article_trainset'
article_market_path='E:/data/knowledge/classify/car/article_trainset_market'
feature_path='E:/data/knowledge/classify/car/trainset_feature'
feature_market_path='E:/data/knowledge/classify/car/trainset_feature_market'
pos_path='E:/data/knowledge/tools/postag'
punc_path='E:/data/knowledge/tools/punctuation'
klword_path='E:/data/knowledge/tools/knowledgeable_word'
train_path='E:/data/knowledge/classify/car/trainset_feature_market'
test_path='E:/data/knowledge/classify/car/testset_feature_market'
corpus = Corpus()
corpus.run(articles_path, article_market_path, \
pos_path, punc_path, klword_path, feature_path, feature_market_path, \
train_path, test_path)
# -*- encoding = gb18030 -*- # package importing start import os import sys sys.path.append(os.path.abspath(os.path.dirname(__file__) + '/../..')) from tag.run import Corpus # package importing end # run this script tag_tree_path = sys.argv[1] sentences_market_path = sys.argv[2] tags_market_path = sys.argv[3] dict_market_path = sys.argv[4] corpus = Corpus() corpus.run_tag_sentences(tag_tree_path, sentences_market_path, tags_market_path, dict_market_path)
# -*- encoding = gb18030 -*- # package importing start import os import sys sys.path.append(os.path.abspath(os.path.dirname(__file__) + '/../..')) from tag.run import Corpus # package importing end # run this script sentences_path = sys.argv[1] sentences_market_path = sys.argv[2] corpus = Corpus() corpus.run_convert_sentences(sentences_path, sentences_market_path)
