Similarity Analysis Using BERT and Jaccard Methods in the Business Processes of Bawaslu RI

Abstract

This study aims to calculate and analyze both semantic and structural similarities in the business processes of the Election Supervisory Board of the Republic of Indonesia (Bawaslu RI). For semantic similarity, the BERT (Bidirectional Encoder Representations from Transformers) method is employed, as it provides contextual representations of words and sentences, thereby aiding in understanding the relationships between entities and elements within business processes. Meanwhile, structural similarity is measured using the Jaccard method due to its simplicity and fairness in evaluating the similarity between process models. This analysis offers deeper insights into workflows, inter-unit interactions, and responsibilities in achieving organizational goals. The results of this study are expected to provide recommendations for improving Bawaslu RI’s business processes to better align with organizational needs at various levels, as well as to enhance effectiveness, transparency, and adaptability to change.