Classifying Indonesian Cyber Crime Cases under ITE Law Using a Hybrid of Mutual Information and Support Vector Machine

In Indonesia, the process of identifying and categorizing cyberlaw infringements traditionally involves manual procedures administered by experts, lawyers, or law enforcement personnel. This study introduces a method to enhance the analysis and processing of case chronological data through the application of text mining. Using the Support Vector Machine for classification, alongside feature extraction both with and without Mutual Information, the study aims to automate the classification of cybercrime cases. The preprocessing phase encompasses text cleaning, case folding, stop word removal, stemming, and tokenization and weighting with TF-IDF. The model achieved an accuracy rate of 95.45% during evaluation and 91.42% when tested on 35 data points with 1500 selected features. This performance surpasses the classification accuracy obtained in previous research.