Automated Text Summarization and Topic Detection on News Aggregation System Using BART and SVM

Abstract

With a large amount of news consumed by the public, it is impossible to digest all the available news. This paper developed an automated text summarization and topic detection algorithm for news articles, allowing the public to read summarized news without losing the essential points of the news. The algorithm will then be used to build and develop a system that has news aggregation technology. First, the system will scrape news articles from various sources, then topic detection and text summarization will be applied to each article before finally being displayed. The methodology used in this research can be divided into data gathering, topic detection, text summarization, and system development. The result of this research shows that the Support Vector Machine performed exceptionally well in topic detection tasks, better than other supervised learning algorithms used in this research, whereas Bidirectional and Auto-Regressive Transformer (BART) with the appropriate parameters performed relatively well in text summarization. To conclude, topic detection and automated text summarization can both be combined and used to develop a news aggregation system, with Support Vector Machine and BART both performing well in their respective tasks.