def slashtags_eval(goldpath, testpath, delimiter, out_f=sys.stdout, tagmap=None, matrix=False, details=False, length_limit=None): ''' Evaluate a "slashtags" format file :param goldpath: :type goldpath: :param testpath: :type testpath: :param delimiter: :type delimiter: :param out_f: :type out_f: :param tagmap: :type tagmap: :rtype: POSEvalDict ''' gold_c = POSCorpus.read_slashtags(goldpath) test_c = POSCorpus.read_slashtags(testpath) poseval(test_c, gold_c, out_f, matrix=matrix, details=details, length_limit=length_limit)
def slashtags_to_simpletagger(in_path, out_path): p = POSCorpus.read_slashtags(in_path) out_f = open(out_path, 'w', encoding='utf-8') for inst in p: sf = SequenceFeature(inst) while sf: out_f.write('%s ' % sf.form) # out_f.write('word-%s ' % sf.form) # out_f.write('pre-3-%s ' % sf.prefix(3)) # out_f.write('pre-2-%s ' % sf.prefix(2)) # # out_f.write('suf-3-%s ' % sf.suffix(3)) # out_f.write('suf-2-%s ' % sf.suffix(2)) # # #=================================================================== # # Context Features # #=================================================================== # out_f.write('prev-%s ' % sf.prev().form) # out_f.write('next-%s ' % sf.next().form) # # #=================================================================== # # More Context # #=================================================================== # out_f.write('prev-prev-%s ' % sf.prev().prev().form) # out_f.write('next-next-%s ' % sf.next().next().form) # Finally, write out the label out_f.write('%s\n' % sf.label) sf = sf.next() out_f.write('\n') out_f.close()