third_order,
label_set,
size_features
]
all_cost_functions = [
micro_f1_cost,
micro_f1_cost_squared,
micro_f1_cost_plusone,
micro_f1_cost_plusepsilon,
binary_cost,
inverse_micro_f1_cost,
uniform_cost
]
all_extractor_fn_names = get_function_names(all_extractor_fns)
base_extractor_fn_names = get_function_names(base_extractors)
all_cost_fn_names = get_function_names(all_cost_functions)
for ngrams in [1]:
logger.info("*" * LINE_WIDTH)
logger.info("NGRAM SIZE: {ngram}".format(ngram=ngrams))
for stemmed in [True]:
logger.info("*" * LINE_WIDTH)
logger.info("Stemmed: {stemmed}".format(stemmed=stemmed))
# update top level stem setting too
config["stem"] = stemmed
# so the base extractors are the MVP for getting to a basic parser, then additional 'meta' parse
# features from all_extractors can be included
base_extractors = [
single_words, word_pairs, three_words, between_word_features
]
all_extractor_fns = base_extractors + [
word_distance, valency, unigrams, third_order, label_set, size_features
]
all_cost_functions = [
micro_f1_cost, micro_f1_cost_squared, micro_f1_cost_plusone,
micro_f1_cost_plusepsilon, binary_cost, inverse_micro_f1_cost, uniform_cost
]
all_extractor_fn_names = get_function_names(all_extractor_fns)
base_extractor_fn_names = get_function_names(base_extractors)
all_cost_fn_names = get_function_names(all_cost_functions)
for ngrams in [1]:
logger.info("*" * LINE_WIDTH)
logger.info("NGRAM SIZE: {ngram}".format(ngram=ngrams))
for stemmed in [True]:
logger.info("*" * LINE_WIDTH)
logger.info("Stemmed: {stemmed}".format(stemmed=stemmed))
# update top level stem setting too
config["stem"] = stemmed