Newsgist: video generation from news stories

Abstract

Digital transition has started to change the way people read news articles more through a digital device and less on paper. Youngsters today do not spend enough time reading news articles. In this work, a knowledge-driven news story generation using collaborative learning to represent the gist of news is proposed. The entire work focuses on two major concerns. Initially, the dialogues associated with the corresponding speaker are extracted from the news. Secondly, the audio of the mapped dialogues is incorporated into the final video. Logistic Regression is deployed to identify the theme the news.  Deep learning techniques are employed to identify the main characters in a supervised manner  using Named Entity Recognition (NER) tagging algorithm, suitable cartoon dispositions and their semantic relations. This approach improves not the reader's comprehension and creativity but also improves mutual goals, opportunities for peer discussion and engaging the underachievers to think reflexively. In addition, it also improves the learner’s motivation and participation. The proposed framework outperforms an accuracy of 83.98% when compared with conventional methods also suggests that the readers found the packages interesting and informative on digital devices. Moreover, this method can be used efficiently in real-time for various applications.