[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"ssr-repo-dspace000002638":3},{"id":4,"title":5,"authors":6,"files":8,"advisors":35,"authors_detail":36,"advisors_detail":40,"publisher":41,"publish_place":42,"language":43,"type":44,"type_slug":45,"year":46,"department":47,"study_program":48,"keywords":49,"keywords_detail":56,"abstract":69,"abstract_html":70,"url":71},"repo-dspace000002638","Automatic Summarization of Student Assignment Documents to Support Lecturer Assessment Using IndoBERT and K-Means Clustering",[7],"Evan Fadhilah Dzulfikar",[9,13,16,19,23,26,29,32],{"chapter":10,"requires_login":11,"label":12,"available":11},"abstrak",false,"Abstrak",{"chapter":14,"requires_login":11,"label":15,"available":11},"bab1","Bab I",{"chapter":17,"requires_login":11,"label":18,"available":11},"bab2","Bab II",{"chapter":20,"requires_login":21,"label":22,"available":11},"bab3",true,"Bab III",{"chapter":24,"requires_login":21,"label":25,"available":11},"bab4","Bab IV",{"chapter":27,"requires_login":21,"label":28,"available":11},"bab5","Bab V",{"chapter":30,"requires_login":21,"label":31,"available":11},"full","Dokumen Lengkap",{"chapter":33,"requires_login":11,"label":34,"available":21},"open","Akses Terbuka",[],[37],{"name":7,"nim":38,"slug":39},"2141720167","evan-fadhilah-dzulfikar",[],"Politeknik Negeri Malang","Malang","Inggris","Skripsi (D4)","skripsi-d4",2025,"D-IV TEKNIK INFORMATIKA",null,[50,51,52,53,54,55],"BERT","Deep Learning","K-Means","Machine Learning","Assesment","Text Summarization",[57,59,61,63,65,67],{"name":50,"slug":58},"bert",{"name":51,"slug":60},"deep-learning",{"name":52,"slug":62},"k-means",{"name":53,"slug":64},"machine-learning",{"name":54,"slug":66},"assesment",{"name":55,"slug":68},"text-summarization","Assessing student assignments is a crucial process in higher education, yet it often consumes considerable time, especially when dealing with long and complex documents. This study proposes an automatic summarization system for Indonesian-language student assignments to assist lecturers in the assessment process. The system is designed by implementing IndoBERT embedding and the KMeans clustering algorithm. The process begins with data preprocessing, including text cleaning and sentence extraction. Each sentence is then converted into a vector using IndoBERT, followed by dimensionality reduction using Principal Component Analysis (PCA) to enhance clustering efficiency. The sentence embeddings are grouped using K-Means, and one sentence closest to each cluster centroid is selected as part of the final summary. The system was evaluated in two aspects: summary quality and scoring accuracy. Evaluation using the ROUGE metric shows that the system achieved a ROUGE-1 score of 0.7300 (recall), 0.5952 (precision), and 0.6348 (F1-score) at the optimal configuration (k=13). Meanwhile, the automatic scoring evaluation yielded a Pearson correlation of –0.23, a Mean Absolute Error (MAE) of 16.96, and a Root Mean Square Error (RMSE) of 20.29. These results indicate that the system is capable of generating reasonably representative summaries, but the accuracy of the automatic scoring remains limited and requires further refinement to align more closely with manual assessments.","\u003Cp>Assessing student assignments is a crucial process in higher education, yet it often consumes considerable time, especially when dealing with long and complex documents. This study proposes an automatic summarization system for Indonesian-language student assignments to assist lecturers in the assessment process. The system is designed by implementing IndoBERT embedding and the KMeans clustering algorithm. The process begins with data preprocessing, including text cleaning and sentence extraction. Each sentence is then converted into a vector using IndoBERT, followed by dimensionality reduction using Principal Component Analysis (PCA) to enhance clustering efficiency. The sentence embeddings are grouped using K-Means, and one sentence closest to each cluster centroid is selected as part of the final summary. The system was evaluated in two aspects: summary quality and scoring accuracy. Evaluation using the ROUGE metric shows that the system achieved a ROUGE-1 score of 0.7300 (recall), 0.5952 (precision), and 0.6348 (F1-score) at the optimal configuration (k=13). Meanwhile, the automatic scoring evaluation yielded a Pearson correlation of –0.23, a Mean Absolute Error (MAE) of 16.96, and a Root Mean Square Error (RMSE) of 20.29. These results indicate that the system is capable of generating reasonably representative summaries, but the accuracy of the automatic scoring remains limited and requires further refinement to align more closely with manual assessments.\u003C\u002Fp>","https:\u002F\u002Frepository.polinema.ac.id\u002Fcollection\u002Fdetail?id=dspace000002638&year=2025&title=automatic-summarization-of-student-assignment-documents-to-support-lecturer-assessment-using-indobert-and-k-means-clustering"]