def ESA_1(fn="output/CV_Gab2005_STOP_CATS.json"):
_log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO, format='%(name)s: %(levelname)-8s %(message)s')
cnt_articles, article_ids, train_set_dict = t.process_hgw_xml()
_log.info("*** Data loaded. ***")
while True:
# enter 1 if you want tfidf processing to be performed, else saving the data will be only done
s = raw_input("Press enter to start a process cycle:\n")
try:
reload(t)
except NameError:
_log.error("Could not reload the module.")
try:
if int(s) == 1:
M, N, CSC_matrix, word_index = t.tfidf_normalize(cnt_articles, article_ids, train_set_dict)
_log.info("*** Tfidf calculation successfully done. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
try:
t.save_CV_all(CSC_matrix, fn, word_index, article_ids)
_log.info("*** Data successfully saved. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
_log.info("Cycle ready.\n")
def ESA():
global F_IN
global F_OUT_CV
global F_OUT_AID
_log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO, format="%(name)s: %(levelname)-8s %(message)s")
articles_with_aid = r.read_in_wikiextractor_output(F_IN, F_OUT_AID)
# print r.articles_selected(777)
_log.info("*** Data loaded. ***")
tf_idf_matrix, word_index, long_articles = t.tfidf_normalize(articles_with_aid)
s.save_CV_all(tf_idf_matrix, word_index, long_articles, F_OUT_CV)
while True:
# enter 1 if you want tfidf processing to be performed, else saving the data will be only done
c = raw_input("Press enter to start a process cycle:\n")
try:
reload(t)
except NameError:
_log.error("Could not reload the module.")
try:
if int(c) == 1:
tf_idf_matrix, word_index, long_articles = t.tfidf_normalize(articles_with_aid)
_log.info("*** Tfidf calculation successfully done. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
try:
try:
reload(s)
except NameError:
_log.error("Could not reload the module.")
if int(c) == 2:
s.save_CV_all(tf_idf_matrix, word_index, long_articles, F_OUT_CV)
_log.info("*** Data successfully saved. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
_log.info("Cycle ready.\n")
def ESA_selected():
_log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO, format='%(name)s: %(levelname)-8s %(message)s')
cnt_articles, article_ids, train_set_dict = t.process_hgw_xml(selected=True, f_in = "Gabrliovich_preprocessed/20051105_pages_articles.hgw.xml",f_articles_out = "output/AID_titles_test_selected.tsv")
_log.info("*** Data loaded. ***")
while True:
# enter 1 if you want tfidf processing to be performed, else saving the data will be only done
s = raw_input("Press enter to start a process cycle:\n")
try:
reload(t)
except NameError:
_log.error("Could not reload the module.")
try:
if int(s) == 1:
M, N, CSC_matrix, word_index = t.tfidf_normalize(cnt_articles, article_ids, train_set_dict)
_log.info("*** Tfidf calculation successfully done. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
try:
if int(s) == 2:
fn="output/CV_Gab2005_true_selected_by_Gab.json"
t.save_CV_all(CSC_matrix, fn, word_index, article_ids)
_log.info("*** Data successfully saved. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
try:
if int(s) == 3:
fn="output/CV_Gab2005_true_selected_by_Gab_pruned.json"
t.save_CV_with_sliding_window_pruning(CSC_matrix, fn, word_index, article_ids)
_log.info("*** Data successfully saved. ***")
except:
e = sys.exc_info()[0]
_log.error("Caught exception from the process\n%s\n%s" % (e, traceback.format_exc()))
_log.info("Cycle ready.\n")
