def get_model_similarities_results(self, questions, number_of_models):
name_array = []
not_wanted = ['npy', 'Readme.md']
onlyfiles = [f for f in listdir(os.path.dirname(os.path.realpath(__file__)) + "/LeaningAlgoImpl/Models") if
isfile(join(os.path.dirname(os.path.realpath(__file__)) + "/LeaningAlgoImpl/Models", f))]
for file_index in range(0, len(onlyfiles)):
if (not_wanted[0] in onlyfiles[file_index] or not_wanted[1] in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
# print(name_array)
random.shuffle(name_array)
results = []
i = 0
for name in name_array:
if number_of_models > i:
print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) + "/LeaningAlgoImpl/Models/" + name)
results.append(finished_model.model_similarity_results(questions))
i += 1
else:
continue
# print(results)
return results
def get_model_similarities_results():
questions = "wordsim353.tsv"
name_array = []
not_wanted = ['npy', 'Readme.md']
onlyfiles = [
f for f in listdir(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models") if isfile(
join(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models", f))
]
for file_index in range(0, len(onlyfiles)):
if (not_wanted[0] in onlyfiles[file_index]
or not_wanted[1] in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
# print(name_array)
results = []
for name in name_array:
print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models/" + name)
pearson, spearman, oov = finished_model.human_similarity_test(
questions)
fixed_result = [
name, 10, spearman[0], spearman[1], pearson[0], pearson[1]
]
print(fixed_result)
results.append(fixed_result)
print(results)
for res in results:
print(res)
if (os.path.isfile("individual_models_human_similarity_english.csv")):
f = open("individual_models_human_similarity_english.csv", "a")
else:
f = open("individual_models_human_similarity_english.csv", "w")
np.savetxt(f, [res], delimiter=',', newline="\n", fmt="%s")
f.close()
# print(results)
return results
def get_multiple_certainess_tie_handling_results(self, questions,
number_of_models, topn):
name_array = []
not_wanted = 'npy'
onlyfiles = [
f for f in listdir(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models") if isfile(
join(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models", f))
]
for file_index in range(0, len(onlyfiles)):
if (not_wanted in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
#print(name_array)
models = []
i = 0
for name in name_array:
if number_of_models > i:
#print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models/" + name)
models.append(finished_model)
i += 1
else:
continue
#print(models)
results = []
for model in models:
results.append(model.get_acc_results_extra(topn, questions))
#print(results)
return results
def get_models_clustering_results(self, test_set, number_of_models):
name_array = []
not_wanted = ['npy', 'Readme.md']
onlyfiles = [
f for f in listdir(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models") if isfile(
join(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models", f))
]
for file_index in range(0, len(onlyfiles)):
if (not_wanted[0] in onlyfiles[file_index]
or not_wanted[1] in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
# print(name_array)
random.shuffle(name_array)
results = []
i = 0
for name in name_array:
if number_of_models > i:
print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models/" + name)
# If model knows less than 3 words we ignore it!
try:
results.append(finished_model.get_model_clusters(test_set))
except IndexError:
print("model knows to little")
i += 1
else:
continue
# print(results)
return results
import LeaningAlgoImpl.Finished_Models as FM
import logging, os, time
logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.INFO)
dir_path = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
finished_model = FM.Finished_Models()
finished_model.get_model(dir_path + '/Code/LeaningAlgoImpl/Models/CBOW,0,5,0,10,100,1,90000_Trained_on1000000articles')
print(finished_model)
finished_model.acc()#'danish-topology.txt')
print(finished_model.acc('questions-words.txt'))
print("DANISH")
#print(finished_model.danish_acc('questions-words.txt'))
vocab = finished_model.get_vocabulary()
if 'en' in vocab:
print('hej')
print(finished_model.danish_acc('danish-topology.txt'))
print(finished_model.special_danish_acc('special-danish-topology.txt'))
#finished_model.get_vocabulary()
#print('similarity')
#print(finished_model.similarity('en', 'er'))
def get_model_clustering_results():
name_array = []
not_wanted = ['npy', 'Readme.md']
onlyfiles = [
f for f in listdir(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models") if isfile(
join(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models", f))
]
for file_index in range(0, len(onlyfiles)):
if (not_wanted[0] in onlyfiles[file_index]
or not_wanted[1] in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
# print(name_array)
for test in [
"Navneord-udsagnsord-tillægsord.csv", "Frugt-dyr-køretøjer.csv",
"Hus-værktøj-kropsdele.csv"
]:
dir_path = os.path.dirname(os.path.dirname(os.path.realpath(__file__)))
test_set = dir_path + '/Code/TestingSet/' + test
reals = []
with open(test_set) as csvfile:
test_reader = csv.reader(csvfile, delimiter=',')
# initialize first cluster
cluster = []
for row in test_reader:
if not row:
# Add new cluster
reals.append(cluster)
cluster = []
else:
cluster.append(''.join(row))
# add last cluster
reals.append(cluster)
print(reals)
results = []
for name in name_array:
print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models/" + name)
try:
result = finished_model.clustering(reals)
print(result)
fixed_result = [name, result[1], result[3]]
print(fixed_result)
results.append(fixed_result)
except IndexError:
print("model knows to little")
print(results)
for res in results:
print(res)
if (os.path.isfile("individual_models_clustering_english" + test +
".csv")):
f = open(
"individual_models_clustering_english" + test + ".csv",
"a")
else:
f = open(
"individual_models_clustering_english" + test + ".csv",
"w")
np.savetxt(f, [res], delimiter=',', newline="\n", fmt="%s")
f.close()
# print(results)
return results
def get_multiple_results(topn=1):
questions = "questions-words.txt"
name_array = []
not_wanted = ['npy', 'Readme.md']
onlyfiles = [
f for f in listdir(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models") if isfile(
join(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models", f))
]
for file_index in range(0, len(onlyfiles)):
if (not_wanted[0] in onlyfiles[file_index]
or not_wanted[1] in onlyfiles[file_index]):
continue
else:
name_array.append(onlyfiles[file_index])
print(name_array)
reals = get_expected_acc_results(questions)
results = []
for name in name_array:
print(name)
finished_model = FM.Finished_Models()
finished_model.get_model(
os.path.dirname(os.path.realpath(__file__)) +
"/LeaningAlgoImpl/Models/" + name)
#models.append(finished_model)
res = finished_model.get_acc_results(1, questions)
correct = []
wrong = []
number_of_correct = 0
number_of_wrong = 0
for j in range(0, len(res)):
if res[j] is not None:
#print(res[j][0])
predicted = res[j][0]
else:
predicted = res[j]
expected = reals[j]
if predicted == expected:
correct_message = ("predicted: %s correct" % (predicted))
correct.append(correct_message)
number_of_correct += 1
else:
wrong_message = ("predicted: %s, should have been: %s" %
(predicted, expected))
wrong.append(wrong_message)
number_of_wrong += 1
results.append([name, len(correct), len(wrong)])
print(reals)
print(results)
print(len(name_array))
print(len(results))
for res in results:
print(res)
if (os.path.isfile("individual_models_acc_english.csv")):
f = open("individual_models_acc_english.csv", "a")
else:
f = open("individual_models_acc_english.csv", "w")
np.savetxt(f, [res], delimiter=',', newline="\n", fmt="%s")
f.close()
# print(results)
return results