def get_article_representation(self, article_title):
''' Returns a bag of words build from the text content
on the given article.
@return: a list of tokens '''
article_text = wikipedia_api_util.query_page_content_text(article_title) # get the text of the article's page
cleaned_tokens = text_util.get_clean_BOW_doc(article_text) # clean the article text
return cleaned_tokens
def get_article_representation(self, article_title):
''' Returns a bag of words build from the title
of the categories of the given article.
@return: a list of tokens '''
category_titles = wikipedia_api_util.query_categories_of_res(article_title)
category_titles_str = self.__format_category__(' '.join(category_titles))
cleaned_titles = text_util.get_clean_BOW_doc(category_titles_str)
return cleaned_titles
def get_article_representation(self, article_title):
''' Returns a bag of words build from the title of the given article.
@return: a list of tokens '''
article_title_str = article_title.replace('_', ' ')
cleaned_title = text_util.get_clean_BOW_doc(article_title_str)
return cleaned_title
