default=True)
input_flags = parser.parse_args()
print(input_flags)
KNN_similarity_to_report_list = [
"cosine", "dice", "jaccard", "asymmetric", "tversky"
]
dataset_list = ["movielens100k"]
for dataset_name in dataset_list:
read_data_split_and_search(
dataset_name,
flag_baselines_tune=input_flags.baseline_tune,
flag_DL_article_default=input_flags.DL_article_default,
flag_print_results=input_flags.print_results,
)
if input_flags.print_results:
generate_latex_hyperparameters(
result_folder_path="result_experiments/{}/".format(
CONFERENCE_NAME),
algorithm_name=ALGORITHM_NAME,
experiment_subfolder_list=dataset_list,
ICM_names_to_report_list=["ICM_genre"],
other_algorithm_list=[MCRecML100k_RecommenderWrapper],
KNN_similarity_to_report_list=KNN_similarity_to_report_list,
split_per_algorithm_type=True)
parser.add_argument('-b', '--baseline_tune', help="Baseline hyperparameter search", type = bool, default = False)
parser.add_argument('-a', '--DL_article_default', help="Train the DL model with article hyperparameters", type = bool, default = False)
parser.add_argument('-p', '--print_results', help="Print results", type = bool, default = True)
parser.add_argument('-m', '--MF_baseline_tune', help="Matrix Factorization hyperparameter search", type = bool, default = False)
input_flags = parser.parse_args()
print(input_flags)
KNN_similarity_to_report_list = ["cosine", "dice", "jaccard", "asymmetric", "tversky"]
dataset_list = ["movielens20m", "netflixPrize"]
for dataset_name in dataset_list:
read_data_split_and_search(dataset_name,
flag_baselines_tune=input_flags.baseline_tune,
flag_MF_baselines_tune = input_flags.MF_baseline_tune,
flag_DL_article_default= input_flags.DL_article_default,
flag_print_results = input_flags.print_results,
)
if input_flags.print_results:
generate_latex_hyperparameters(result_folder_path ="result_experiments/{}/".format(CONFERENCE_NAME),
algorithm_name= ALGORITHM_NAME,
experiment_subfolder_list = ["{}_cold_user".format(dataset) for dataset in dataset_list],
other_algorithm_list = [Mult_VAE_RecommenderWrapper],
KNN_similarity_to_report_list = KNN_similarity_to_report_list,
split_per_algorithm_type = True)
KNN_similarity_to_report_list = [
"cosine", "dice", "jaccard", "asymmetric", "tversky"
]
dataset_list = [
'movielens1m_original', 'movielens1m_ours', 'tafeng_original',
'tafeng_ours'
]
for dataset_name in dataset_list:
read_data_split_and_search(
dataset_name,
flag_baselines_tune=input_flags.baseline_tune,
flag_DL_article_default=input_flags.DL_article_default,
flag_print_results=input_flags.print_results,
)
if input_flags.print_results:
generate_latex_hyperparameters(
result_folder_path="result_experiments/{}/".format(
CONFERENCE_NAME),
algorithm_name=ALGORITHM_NAME,
experiment_subfolder_list=dataset_list,
ICM_names_to_report_list=["ICM_all", "ICM_original"],
UCM_names_to_report_list=["UCM_all"],
other_algorithm_list=[
DeepCF_RecommenderWrapper, CoupledCF_RecommenderWrapper
],
KNN_similarity_to_report_list=KNN_similarity_to_report_list,
split_per_algorithm_type=True)
dataset_variant_list = ["a", "t"]
train_interactions_list = [1, 10]
for dataset_variant in dataset_variant_list:
for train_interactions in train_interactions_list:
read_data_split_and_search(
dataset_variant,
train_interactions,
flag_baselines_tune=input_flags.baseline_tune,
flag_DL_article_default=input_flags.DL_article_default,
flag_print_results=input_flags.print_results,
)
if input_flags.print_results:
generate_latex_hyperparameters(
result_folder_path="result_experiments/{}/".format(
CONFERENCE_NAME),
algorithm_name=ALGORITHM_NAME,
experiment_subfolder_list=[
"citeulike_{}_{}".format(dataset_variant, train_interactions)
for dataset_variant in dataset_variant_list
for train_interactions in train_interactions_list
],
ICM_names_to_report_list=["ICM_tokens_TFIDF", "ICM_tokens_bool"],
KNN_similarity_to_report_list=KNN_similarity_to_report_list,
other_algorithm_list=[CollaborativeDL_Matlab_RecommenderWrapper],
split_per_algorithm_type=True)
read_data_split_and_search(
dataset_name,
cold_start=input_flags.cold_start,
flag_baselines_tune=input_flags.baseline_tune,
flag_DL_article_default=input_flags.DL_article_default,
flag_DL_tune=input_flags.DL_tune,
flag_print_results=input_flags.print_results)
# mantain compatibility with latex parameteres function
if input_flags.cold_start and input_flags.print_results:
for n_cold_item in cold_start_items_list:
generate_latex_hyperparameters(
result_folder_path="result_experiments/{}/".format(
CONFERENCE_NAME),
algorithm_name="{}_cold_{}".format(ALGORITHM_NAME,
n_cold_item),
experiment_subfolder_list=dataset_cold_start_list,
other_algorithm_list=other_algorithm_list,
KNN_similarity_to_report_list=KNN_similarity_to_report_list,
split_per_algorithm_type=True)
elif not input_flags.cold_start and input_flags.print_results:
generate_latex_hyperparameters(
result_folder_path="result_experiments/{}/".format(
CONFERENCE_NAME),
algorithm_name=ALGORITHM_NAME,
experiment_subfolder_list=dataset_list,
other_algorithm_list=other_algorithm_list,
KNN_similarity_to_report_list=KNN_similarity_to_report_list,
split_per_algorithm_type=True)
