Akademik Veri Yönetim Sistemi
9th International Artificial Intelligence and Data Processing Symposium, IDAP 2025, Malatya, Türkiye, 6 - 07 Eylül 2025, (Tam Metin Bildiri)
The increase of written data on the Internet and the development of transformer-based natural language processing models have accelerated work in the fields of text generation and summarization. Text summarization is a process that aims to convey meaningful parts of relatively large content in a shorter form. Text summarization is generally divided into two parts: extractive and abstractive. In this study, we create abstract summaries of news articles using transformer-based models such as BART and MT5, applying them to Turkish data from the MLSUM dataset, which contains online news content in multiple languages. We also tweaked the MT5 model to boost its performance. The success of the models was evaluated with ROUGE and BERTScore scores, which are widely used in the field of text summarization, and the average summary generation time of each model was calculated. In line with the results obtained, it was observed that the performance of the models trained with domain-specific Turkish data was more successful than the other models in summarization, and it was aimed to contribute to the Turkish text summarization studies.