def create_speeches_by_person(input_speeches_file,\
input_subjects_file,\
output_person_input_for_word_cloud_directory,\
output_person_word_cloud):
MalletParser.createInputForMallet()
speeches_dict = read_speeches_by_person(input_speeches_file)
subjects_dict = read_subjects(input_subjects_file)
person_fd_dict = create_key_fd_dict(speeches_dict)
person_subjects = create_person_subjects_dict(subjects_dict, person_fd_dict)
write_persons_output_files(output_person_input_for_word_cloud_directory, output_person_word_cloud, person_subjects)
def create_speeches_by_year(input_speeches_file,\
output_person_input_for_word_cloud_directory,\
output_year_word_cloud):
MalletParser.createInputForMallet()
speeches_dict = read_speeches_by_year(input_speeches_file)
year_fd_dict = create_key_fd_dict(speeches_dict)
year_frequent_words_dict = {}
tf_idf = calculate_tf_idf(speeches_dict)
for key, value in year_fd_dict.items():
year_frequent_words_dict[key] = get_most_common_words(value, tf_idf)
write_by_year_output_files(output_person_input_for_word_cloud_directory, output_year_word_cloud, year_frequent_words_dict)
