import test from obo import read_terms from onto_utils import read_heads, read_blacklist, read_types, read_lemma # First read_blacklist(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/blacklist.txt')) # LEARNING read_terms(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/bacteria_habitat_OntoBiotope-34')) read_heads(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/heads_tolearn_onto'), action='learn') #test.cleaning_helper() print '#' * 100 #test.cleaning_helper() # LEARNING print 'LEARNING' print "Reading lemma" read_lemma(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/lemma')) print "Reading types" read_types(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/types')) print "Reading heads" read_heads(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/heads_tolearn_dico'), action='learn') print "Saving" test.save(prefix='/bibdev/travail/typage/typage_biotope_task3.4/dumps/expe1_20120912/expe1_20120912_after_learning') # TAGGING print "\nTAGGING" print "Tagging heads" read_heads(open('/bibdev/travail/typage/typage_biotope_task3.4/data/expe_20120912/heads_totag'), action='tag') print "Saving" test.save(prefix='/bibdev/travail/typage/typage_biotope_task3.4/dumps/expe1_20120912/expe1_20120912_after_tagging')
if BLACKLIST_OK: print 'Reading blacklist' read_blacklist(codecs.open(BASE_DATA + u'blacklist', encoding='UTF-8')) print 'Blacklist content: %s\n' % ', '.join(test.blacklist) print '-' * 80 if ONTO_OK: ############################################################################# # LEARNING ONTO print 'LEARNING ONTO' print '-' * 80 ############################################################################# print "Reading onto" read_terms(codecs.open(BASE_DATA + u'onto')) print '%d terms currently in memory.\n' % len(test.terms) print 'The following inconsistencies were found in the ontology:' test.cleaning_helper() print print "Reading onto heads" read_heads(codecs.open(BASE_DATA + u'heads_tolearn_onto', encoding='UTF-8'), action='learn') print '%d terms currently in memory.\n' % len(test.terms) print 'The following inconsistencies were found in the ontology:' test.cleaning_helper() print print "Saving after learning onto" test.save(prefix=BASE_DUMPS + EXPE + u'_after_learning_onto') print '-' * 80