Automatic Summarization of Court Decision Documents over Narcotic Cases Using BERT

Reviewing court decision documents for references in handling similar cases can be time-consuming. From this perspective, we need a system that allow the summarization of to enable adequate information extraction. This study used 50 taken from official website Supreme Court Republic Indonesia, with raised being Narcotics and Psychotropics. The document dataset was divided into two types, identity defendant without defendant's identity. We BERT specific IndoBERT model summarize documents. uses four types IndoBert models: IndoBERT-Base-Phase 1, IndoBERT-Lite-Bas-Phase IndoBERT-Large-Phase IndoBERT-Lite-Large-Phase 1. also three ratios ROUGE-N summarizing consisting 20%, 30%, 40% ratios, as well ROUGE1, ROUGE2, ROUGE3. results have found pre-trained had better performance or ratio. highest ROUGE score produced by INDOBERT-LITE-BASE PHASE 1 value 1.00 0.970 at ratio R-1. For future research, it is expected able use other Bert models such Phase-2, LegalBert, etc.