News Curation, Abstract, and Recommender App using Deep Learning Attention Models
Nikit Periwal, Niranjan Mahesh, Namrita Kaur, Nirgund Manavendra P Jayaram, Asha Rani K P, Saahithya Gowrishankar · 2022 International Conference on Edge Computing and Applications (ICECAA) · 2022
It is important to keep ourselves updated on the latest news. People skip reading the news because of the lack of time or if the news articles are too lengthy. The aforementioned issue is addressed by making the news articles more concise and the interests of the user can be catered to by providing them with a personalized news feed. Existing applications either personalize the news feed or provide abstract news. The proposed system aims to provide abstract news and personalize the news feed. For the proposed system, the main components are collecting and extracting the news, news abstraction, recommender system, and the application backend. To collect and extract the contents of the articles, NEWS API and newspaper3k Library are employed. Abstract of the news article is created by employing the pre-trained BART model and for categorizing the news, the DistilBERT model which can categorize an article within 4ms has been fine-tuned. A content-based approach has been used to personalize the news feed. Application backend tracks the data required for personalizing the news and retrieves the recommended articles for the news feed. The proposed system can deliver personalized, curated, and abstract news to each user.