<![CDATA[Advances in information technology have led to a growing volume of news available in digital media. The sheer volume of information often makes it difficult for readers to extract the key points from multiple news articles on the same topic. Therefore, a system is needed that can summarize multiple news documents into a shorter, more easily understandable summary. This study aims to design and implement a multi-document news summarization system using a graph-based approach. The summarization process involves several stages: text preprocessing, weighting using the Term Frequency–Inverse Document Frequency (TF-IDF) method, calculating sentence similarity using cosine similarity, and constructing a graph of relationships between sentences. Subsequently, sentence ranking is performed using the TextRank and LexRank algorithms to determine the most representative sentences for the summary. System evaluation was performed using the Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metric with news documents as pseudo-references. The test results showed a ROUGE-1 score of 0.47 and a ROUGE-L score of 0.64, indicating that the system is capable of generating summaries that adequately represent the important information from the news documents.]]>
Copyrights © 2026
