Text Summarization Berita Online Menggunakan Named Entity Recognition (NER) dan Part of Speech (POS) Tagging

Abstract

The objective of this research is to develop an automated online news system using Named Entity Recognition (NER) and several voice-based technologies. This system will help readers quickly and efficiently grasp core information by extracting key message texts. The methods employed include data collection preparation (case folding, tokenization, stopword removal, filtering, stemming) from CNN Indonesia news pages, as well as utilizing NER and POS tagging to identify entity structures and important statements. Summary Quality Assessment was conducted using the BERTScore metric. The results indicate that the combination of NER and POS Tagging yields the best performance, with an F1-score of 0.74, surpassing the use of NER (0.709) or POS Tagging (0.734) individually. This combination improves precision (0.732) without affecting recall (0.7486), demonstrating its effectiveness in gathering essential information. The study proves that integrating NER and POS tags is effective for text summarization and can serve as a solution for presenting concise and relevant information to users.