def evaluate_genia_dataset():
import logging.config
logging.config.fileConfig(os.path.join('..', 'logging.conf'))
import multiprocessing
from jgtextrank import keywords_extraction_from_tagged_corpus
import time
reader = WordListCorpusReader('../resource', 'smart-stop-list.txt')
stop_list = reader.words()
gs_terms = load_genia_gs_terms('genia_gs_terms.txt')
corpus_directory = os.path.join('GENIAcorpus302', 'text', 'files')
pre_processed_corpus = pre_processing_corpus_with_spacy(corpus_directory,
encoding="utf-8",
lemma=True)
print("term extraction from spaCy pre-processed corpus [%s] ..." %
corpus_directory)
pre_processed_corpus = list(pre_processed_corpus)
start = time.time()
genia_keywords, genia_top_vertices = keywords_extraction_from_tagged_corpus(
pre_processed_corpus,
window=2,
top_p=1,
stop_words=stop_list,
weight_comb="norm_max",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", genia_keywords[:50])
evaluate_results(genia_keywords, gs_terms, "genia+pagerank")
start = time.time()
genia_gcvalue_keywords, genia_gcvalue_top_vertices = keywords_extraction_from_tagged_corpus(
pre_processed_corpus,
window=2,
top_p=1,
stop_words=stop_list,
weight_comb="gcvalue",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", genia_gcvalue_keywords[:50])
evaluate_results(genia_gcvalue_keywords, gs_terms,
"genia+pagerank+gcvalue")
def evaluate_hulth2003_testset():
import logging.config
logging.config.fileConfig(os.path.join('..', 'logging.conf'))
import multiprocessing
from jgtextrank import keywords_extraction_from_tagged_corpus
import time
reader = WordListCorpusReader('../resource', 'smart-stop-list.txt')
stop_list = reader.words()
hulth2003_corpus_directory = os.path.join('Hulth2003', 'Test')
hulth2003_gs_terms = load_Hulth2003_gs_terms(hulth2003_corpus_directory)
hulth2003_corpus_directory = os.path.join('Hulth2003', 'Test')
pre_processed_hulth2003_corpus = pre_processing_corpus_with_spacy(hulth2003_corpus_directory,
encoding="utf-8", lemma=True, default_file_suffix=".abstr")
print("term extraction from spaCy pre-processed hulth2003 corpus [%s] ..." % hulth2003_corpus_directory)
pre_processed_hulth2003_corpus = list(pre_processed_hulth2003_corpus)
start = time.time()
hulth2003_keywords, hulth2003_top_t_vertices = keywords_extraction_from_tagged_corpus(pre_processed_hulth2003_corpus,
window=3,
top_p = 1, stop_words=stop_list,
weight_comb="norm_max",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", hulth2003_keywords[:50])
print("\n")
evaluate_results(hulth2003_keywords, hulth2003_gs_terms, "hulth2003+pagerank")
start = time.time()
hulth2003_keywords, hulth2003_top_t_vertices = keywords_extraction_from_tagged_corpus(pre_processed_hulth2003_corpus,
window=2,
top_p = 1, stop_words=stop_list,
weight_comb="gcvalue",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", hulth2003_keywords[:50])
print("\n")
evaluate_results(hulth2003_keywords, hulth2003_gs_terms, "hulth2003+pagerank+gcvalue")
def evaluate_semeval2017_testset():
import logging.config
logging.config.fileConfig(os.path.join('..', 'logging.conf'))
import multiprocessing
from jgtextrank import keywords_extraction_from_tagged_corpus
import time
gs_terms = load_scienceie_test_dataset("semeval_articles_test",
file_suffix=".ann")
print(len(gs_terms), " gs terms loaded.")
reader = WordListCorpusReader('../resource', 'smart-stop-list.txt')
stop_list = reader.words()
semeval2017_pre_processed_corpus = pre_processing_corpus_with_spacy(
"semeval_articles_test", default_file_suffix=".txt")
semeval2017_pre_processed_corpus = list(semeval2017_pre_processed_corpus)
start = time.time()
semeval2017_keywords, semeval2017_top_t_vertices = keywords_extraction_from_tagged_corpus(
semeval2017_pre_processed_corpus,
window=3,
top_p=1,
stop_words=stop_list,
weight_comb="avg",
export=False,
workers=1)
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", semeval2017_keywords[:50])
print("\n")
evaluate_results(semeval2017_keywords, gs_terms, "pagerank+norm_max")
start = time.time()
semeval2017_gcvalue_keywords, semeval2017_gcvalue_top_t_vertices = keywords_extraction_from_tagged_corpus(
semeval2017_pre_processed_corpus,
window=3,
top_p=1,
stop_words=stop_list,
weight_comb="gcvalue",
export=False,
workers=1)
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", semeval2017_gcvalue_keywords[:50])
print("\n")
evaluate_results(semeval2017_gcvalue_keywords, gs_terms,
"pagerank + GC-Value")
def evaluate_aclrdtec1_dataset():
import logging.config
logging.config.fileConfig(os.path.join('../jgtextrank', 'logging.conf'))
import multiprocessing
from jgtextrank import keywords_extraction_from_tagged_corpus
import time
reader = WordListCorpusReader('../jgtextrank', 'smart-stop-list.txt')
stop_list = reader.words()
# C:\\Data\\NLP-corpus\\ACL RD-TEC\\_all_annotated_candid_term\\_all_annotated_candid_term
aclrdtec1_gs_terms_file_path = os.path.join('/home', 'jieg', 'data',
'ACL RD-TEC-1',
'_all_annotated_candid_term')
print("loading ACLRDTEC 1.0 GS terms from [%s] ..." %
aclrdtec1_gs_terms_file_path)
aclrdtec1_gs_terms = load_aclrdtec1_gs_terms(aclrdtec1_gs_terms_file_path)
print("total [%s] normed GS terms loaded" % len(aclrdtec1_gs_terms))
zipped_corpus_path = os.path.join('/home', 'jieg', 'data', 'ACL RD-TEC-1',
'ACLRDTEC-1.zip')
print("term extraction from spaCy pre-processed corpus [%s] ..." %
zipped_corpus_path)
start = time.time()
doc_content_list = load_all_files_from_zip_file(zipped_corpus_path)
pre_processed_corpus = pre_processing_unzipped_corpus_with_spacy(
doc_content_list)
pre_processed_corpus = list(pre_processed_corpus)
end = time.time()
print("\n")
print("Complete corpus pre-processing. Wall-clock elapsed time: ",
end - start, "s")
print("\n")
start = time.time()
aclrdtec1_keywords, aclrdtec1_top_vertices = keywords_extraction_from_tagged_corpus(
pre_processed_corpus,
window=2,
top_p=1,
stop_words=stop_list,
weight_comb="norm_max",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", aclrdtec1_keywords[:50])
evaluate_results(aclrdtec1_keywords, aclrdtec1_gs_terms,
"aclrdtec1 + pagerank")
start = time.time()
aclrdtec1_gcvalue_keywords, aclrdtec1_gcvalue_top_vertices = keywords_extraction_from_tagged_corpus(
pre_processed_corpus,
window=2,
top_p=1,
stop_words=stop_list,
weight_comb="gcvalue",
export=False,
workers=multiprocessing.cpu_count())
end = time.time()
print("\n")
print("Complete. Wall-clock elapsed time: ", end - start, "s")
print("\n")
print("top 50 keywords: ", aclrdtec1_gcvalue_keywords[:50])
evaluate_results(aclrdtec1_gcvalue_keywords, aclrdtec1_gs_terms,
"aclrdtec1+pagerank+gcvalue")