vocab_size = args.vocab_size
split_ratio = args.split_ratio
print "=================================Preprocess Option Setting================================="
print "\tsaving preprocessed aux path - %s" % aux_path
print "\tsaving preprocessed data path - %s" % data_path
print "\trating data path - %s" % path_rating
print "\tdocument data path - %s" % path_itemtext
print "\tmin_rating: %d\n\tmax_length_document: %d\n\tmax_df: %.1f\n\tvocab_size: %d\n\tsplit_ratio: %.1f" \
% (min_rating, max_length, max_df, vocab_size, split_ratio)
print "==========================================================================================="
R, D_all = data_factory.preprocess(
path_rating, path_itemtext, min_rating, max_length, max_df, vocab_size)
data_factory.save(aux_path, R, D_all)
data_factory.generate_train_valid_test_file_from_R(
data_path, R, split_ratio)
else:
res_dir = args.res_dir
emb_dim = args.emb_dim
pretrain_w2v = args.pretrain_w2v
dimension = args.dimension
lambda_u = args.lambda_u
lambda_v = args.lambda_v
max_iter = args.max_iter
num_kernel_per_ws = args.num_kernel_per_ws
give_item_weight = args.give_item_weight
if res_dir is None:
sys.exit("Argument missing - res_dir is required")
if lambda_u is None:
sys.exit("Argument missing - lambda_u is required")
print("\tsaving preprocessed aux path - %s" % aux_path)
print("\tsaving preprocessed data path - %s" % data_path)
print("\trating data path - %s" % path_rating)
print("\tdocument data path - %s" % path_itemtext)
print("\tprofile data path - %s" % path_usertext)
print ("\tmin_rating: %d\n\tmax_length_document: %d\n\tmax_df: %.1f\n\tvocab_size: %d\n\tsplit_ratio: %.1f" \
% (min_rating, max_length, max_df, vocab_size, split_ratio))
print(
"==========================================================================================="
)
R, D_all = data_factory.preprocess_ext(path_rating, path_itemtext,
path_usertext, min_rating,
max_length, max_df, vocab_size)
data_factory.save(aux_path, R, D_all)
data_factory.generate_train_valid_test_file_from_R(data_path, R,
split_ratio)
else:
res_dir = args.res_dir
emb_dim = args.emb_dim
pretrain_w2v = args.pretrain_w2v
dimension = args.dimension
lambda_u = args.lambda_u
lambda_v = args.lambda_v
max_iter = args.max_iter
num_kernel_per_ws = args.num_kernel_per_ws
give_weight = args.give_weight
threshold_doclen = args.threshold_length_document
threshold_sentlen = args.threshold_length_sentence
binary_rating = args.binary_rating
if res_dir is None:
for i in range(len(movie_ids)):
movie_ids[i] = mapped_ids.index(movie_ids[i])
mapped_ids = sorted(list(set(user_ids)))
for i in range(len(user_ids)):
user_ids[i] = mapped_ids.index(user_ids[i])
filestr = ""
for i in range(len(user_ids)):
writeline = str(user_ids[i]) + "::" + str(movie_ids[i]) + "::" + str(
ratings[i]) + "\n"
filestr += writeline
with open('ratings.dat', 'w') as f:
f.write(filestr)
data = np.array([float(i) for i in ratings])
row = np.array([int(i) for i in user_ids])
col = np.array([int(i) for i in movie_ids])
R = csr_matrix((data, (row, col)))
f = open('test/FS/ratings.all', 'wb')
pickle.dump(R, f)
f.close()
f = open("test/FS/ratings.all", 'rb')
R = pickle.load(f)
f.close()
split_ratio = 0.2
##R = csr_matrix(([int(i) for i in ratings], ([int(i) for i in user_ids], [int(i) for i in movie_ids])))
data_factory.generate_train_valid_test_file_from_R('test/FS/', R, split_ratio)
