Abstract
Many Learning Management Systems (LMS) provide user reviews for training programs, yet these reviews often lack concise summaries, making it difficult for users to evaluate training quality effectively. Therefore, this research purposes to generate abstractive summary insights from large volumes of LMS training reviews. The method used in this research is the Text-to-text Transfer Transformers (T5) algorithm with the T5-Base variant, which is capable of transforming lengthy textual inputs into shorter, coherent summaries. Data collection is scraped from 60 training modules on the LMS and results in over 30,000 review data. The T5-Base model was trained and evaluated using Recall-Oriented Understudy for Gisting Evaluation (ROUGE) metrics, yielding scores of ROUGE-1 is 0.4809, ROUGE-2 is 0.2816, and ROUGE-L is 0.3773, with an overall model accuracy of 75%. These performance indicators confirm that the model is able to capture essential information and produce summaries that effectively reflect the core content of the original reviews. Overall results demonstrate that the T5-Base model effectively produces relevant abstractive summaries, offering a practical solution for improving LMS content transparency and assisting users in selecting suitable training programs. By providing clear and concise summaries, the system improves content transparency and assists users in making informed decisions when selecting training programs. This research highlights the potential of abstractive text summarization to support better information management within modern digital learning environments.