preprocess.py When to run: Add binary vector and form final tag vector for each user Options: --dir: name of the directory containing input data (Default: ‘data’) --input_subdir: name of the subdirectory containing input data (Default: ‘raw’) --output_subdir: name of the subdirectory for output data (Default: ‘preprocessed’) --datetime: datetime to be added to input/output file name (Default: None) Input Files: dir/input_subdir/names_tags_raw_datetime.csv dir/input_subdir/activities_tags_raw_datetime.csv dir/input_subdir/subjects_tags_raw_datetime.csv dir/input_subdir/ user_item_raw_datetime.csv Output Files: dir/output_subdir/names_tags_preprocessed_datetime.csv dir/output_subdir/activities_tags_preprocessed_datetime.csv dir/output_subdir/subjects_tags_preprocessed_datetime.csv dir/output_subdir/ user_item_preprocessed_datetime.csv
RecommendToTopKUser.py When to run: New Activity to notify topK users Options: --dir: name of the directory containing input data (Default: ‘data’) --input_subdir: name of the subdirectory containing input data (Default: ‘preprocessed’) --output_subdir: name of the subdirectory for output data (Default: ‘prediction’) --datetime: datetime to be added to input/output file name (Default: None) --topK: amount of users to recommended per item(Default: 100) Input Files: dir/output_subdir/names_tags_preprocessed_datetime.csv dir/output_subdir/activities_tags_preprocessed_datetime.csv Output Files: dir/output_subdir/Recommend_topK_user_datetime.csv
RecommendToTopKItem.py When to run: Recommend topK activities to new user with less than 3 interactions Options: --dir: name of the directory containing input data (Default: ‘data’) --input_subdir: name of the subdirectory containing input data (Default: ‘preprocessed’) --output_subdir: name of the subdirectory for output data (Default: ‘prediction’) --datetime: datetime to be added to input/output file name (Default: None) --topK: amount of items to recommended per user (Default: 10) Input Files: dir/input_subdir/names_tags_preprocessed_datetime.csv dir/input_subdir/activities_tags_preprocessed_datetime.csv Output Files: dir/output_subdir/Recommend_topK_item_datetime.csv
NeuMF.py When to run: When all users in user_item have more than or equal to 3 interactions with an activity Options: --dir: name of the directory containing input data (Default: ‘data’) --input_subdir: name of the subdirectory containing input data (Default: ‘preprocessed’) --output_subdir: name of the subdirectory for output data (Default: ‘prediction’) --datetime: datetime to be added to input/output file name (Default: None) --topK: amount of items to recommended per user (Default: 10) --epochs: number of epochs to train the model (Default: 100) --batch_size: batch size used to feed to model (Default:256) Input: dir/input_subdir/names_tags_preprocessed_datetime.csv dir/input_subdir/activities_tags_preprocessed_datetime.csv dir/input_subdir/subjects_tags_preprocessed_datetime.csv dir/input_subdir/user_item_preprocessed_datetime.csv Output: dir/output_subdir/Model_topK_Recommendation_datetime.csv